SlideShare a Scribd company logo
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
1
Microsoft SQL Server,
Scalability, &
Database Research
Jim Gray
Researcher
Microsoft Corporation
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
2
Outline
Summary of what you heard. (10 min)
The database scene in general. (10 min)
Scaleability: Farms, Clones,Parts & Packs (15 min)
Microsoft DB research focus. (15 min)
• TerraServer (design and ops).
• RAGS.
• Data Mining
Q&A (10 min)
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
3
Organizations Are Going Online
Building a digital nervous system.
Inexpensive hardware means huge
databases are possible.
But, we are drowning in data.
Databases help organize information.
Microsoft’s goal:
• Information at your fingertips.
• Make it easy to capture,
manage, and
analyze information.
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
4
Microsoft SQL Server 7 Goals
Easy
 Dynamic self management
 Multi-site management
 Operation Scripting
 Job scheduling and execution
 Alert/response management
 Scriptable Install+upgrade
 DBA profiling/tuning tools
 Unicode
 English Language Query
 Integrated with NT Security
 Integrated with NT files
Scalability
Data Warehousing
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
11
Scalability
 Win9x/NTW version
 Dynamic row-level locking
 Improved query optimizer
 Intra-query parallelism
 VLDB improvements
 Replication improvements
 Distributed query
 High Availability Clusters
Easy
Scalability
Data Warehousing
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
12
Scale Down to Windows 95-98
Full function (same as NTW)
Integration with Access 97
MSDE in Office2000 and MSDN
WinCE version demonstrated
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
13
Replication
Transactional and Merge
Remote update
ODBC and OLE DB subscribers
Wizards
Performance
2PC,
RPC
Subscriber
DB2
CICS Subscriber
Subscriber
VSAM
OS 390
DB2
Publisher
Updating Subscriber
(immediate updates)
Distributor
Subscriber
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
14
Query Processor Enhancements
Parallelism
Improved scan, fetch, & sort
Smart hash & merge join
Large joins & grouping
Better query optimization
Multi-index operations
Automatic statistics maintenance
Distributed Query
Heterogeneous Query
Focus on Complex Queries
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
15
•# of emp. per group
•total inc. per group
Local Agg.
4 x 50 rows
+ + + +
Disks
50,000 rows
Global Agg. Result 50 rows
+
Parallel Query
SMP & Disk Parallelism
Plus Distributed
Plus Hash Join (fanciest on the planet)
Plus Optimized Partitioned views
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
16
Distributed Heterogeneous Queries
Data Fusion / Integration
Join spread sheets,
databases,
directories,
Text DBs
etc.
Any source that
exposes OLE DB
interfaces
SQL Server as
gateway,
even on the
desktop
Database
(DB2, VSAM, Oracle, …)
Spreadsheet
Photos
Mail
Maps
Documents
and the Web
Directory
Service
SQL 7.0
Query
Processor
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
17
Utilities
The Key to LARGE Databases
 Backup
• Fuzzy
• Parallel
• Incremental
• Restartable
Recovery
• Fast
• File granularity
Reorganize
• shrinks file
• reclusters file
Auto-Repair
Index creation
~2x faster than 6.5
DBCC
• not required,
• a good practice
• 5x - 100x faster
0
10
20
30
40
50
60
1 2 3 4
SQL Server 6.5
SQL Server 7.0
# of indices
Recovery
time
(
secs
)
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
20
Data Warehousing
 Warehousing Framework
 Visual data modeler
 Microsoft repository
 Data transformation services
(DTS)
 Plato & Dcube - Multi
Dimensional Data Cubes
 English query 2.0
Easy
Scalability
Data Warehousing
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
22
Data Warehouse / Data Analysis
Data Transformation Services
to get data into the warehouse
CUBE (OLE/DB OLAP)
to analyze data
Operational
Data Extact
& Load
Data Warehouse
Storage
OLAP
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
25
Source table
Partition 1
ROLAP
Partition 2
Partition 3
ROLAP
Europe
USA
Asia
MD SQL
SQL
Plato and Data Cube
and HOLAP
“Plato”
server
“Plato”
Designer
Dcube
Client
app
User 1
Dcube
Client
app
User 2
RED
WHITE
BLUE
By Color
By Make & Year
By Color & Year
By Make
By Year
Sum
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
26
English Query
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
27
Easy
Scalable
Data
Warehousing
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
“Shiloh” The Next SQL Server
Shiloh (H1’00) - Strengthen Position
• Data Warehousing leadership
 Materialized Views
 Cascading Referential Integrity
(#1 requested user-group feature)
 XML support
• Scalability
 WinCE support
 W2K VLM (36 and 64 bit)
 Multi-instance support
Yukon – Next Big Step
• Scalability (Clusters, Partitions)
• Programmability
• Ease of Use (Self Tuning, Auto Config)
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
29
Outline
Summary of what you heard. (10 min)
The database scene in general. (10 min)
Scaleability: Farms, Clones,Parts & Packs (15 min)
Microsoft DB research focus. (15 min)
• TerraServer (design and ops).
• RAGS.
• Data Mining
Q&A (10 min)
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
30
Info Capture
You can record everything
you see or hear or read.
What will you do with it?
How will you organize &
analyze it?
Most data will never be seen
Analysis an summarization
are key technologies
Video 8 PB per lifetime (10GBph)
Audio 30 TB (10KBps)
Read or write: 8 GB (words)
See: http://www.lesk.com/mlesk/ksg97/ksg.html
Yotta
Zetta
Exa
Peta
Tera
Giga
Mega
Kilo A Book
.Movi
e
All books
(words)
All Books
MultiMedia
Everything
!
Recorded
A Photo
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
31
Data Tidal Wave
 Seagate 47GB drive @ 783$ (= 1.7 ¢/mb)
• 100 GB penny per MB drive coming in 2000
 10 $/GB = 10 k$/ Terabyte!
• “Everyone” can afford one
 What’s a terror bite?
• If you sell ten billion items a year (e.g Wal-Mart)
• And you record 100 bytes on each one
• Then you get a TeraByte/year
 Where will the terror bytes come from?
• Multimedia (like the TerraServer) and...
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
32
Reducing Data’s Cost-of-Ownership
Self-Managing data
Cost of ownership:
One admin/TB (100K$ vs 10K$)
Admin cost exceeds storage cost.
SQL 7:
Suggests indices
Migrates data away from end of file
Truncates file
Someday:
Automatic move files to balance disks
Online defragmentation & restructuring
Online physical redesign
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
33
OBJECT RELATIONAL
The Next Great DBMS Wave
All DB vendors have added objects to DB
Microsoft is adding DBs to Objects
Integration with COM+
Gives user-defined types and objects
Plug-ins will be Billion dollar industry
• Blades for SQL Server razor
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
34
Why Is XML Important?
Self-describing data
“ABC47-Z”, “100”, “STL”, “C”, “3”, “28”
Data stream in a typical interface…
<INVENTORY>
<PART_NUM>ABC47-Z</PART_NUM>
<QUANTITY>100</QUANTITY>
<WAREHOUSE>STL</WAREHOUSE>
<ZONE>C</ZONE>
<AISLE>3</AISLE>
<BIN>28</BIN>
</INVENTORY>
Same data stream in XML…
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
35
table.xsl
bar.xsl
art.xsl
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
36
XML Applications
Exposing Software as a “Service”
• Websites without UI’s
• Exposed services with common scheme
• Integration points at the enterprise, value-
chain, workgroup, desktop and intelligent
gizmo “levels”
B2B value chains
• Uses XML to transmit wide range of date to a
broad set of stakeholders (regulatory
agencies, suppliers, customers, etc.).
• Leverage for prior efforts like EDI
• BizTalk a key industry effort in this regard
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
37
XML Message
XML
Document
XML Message
XML: BizTalk Framework
Order Processing
MVS CICS
SAP R/3
XML
XML
Service
Interface
Browser
Client Apps
New Form Factors
XML
Document
Another Service
JD Edwards
XML
XML
Library
www.biztalk.org
XML
XML
schema
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
38
Outline
Summary of what you heard. (20 min)
The database scene in general. (10 min)
Scaleability: Farms, Clones,Parts & Packs (10 min)
Microsoft DB research focus. (15 min)
• TerraServer (design and ops).
• RAGS.
• Data Mining
Q&A (15 min)
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
39
Terminology for scaleability
Farms of servers:
• Clones: identical
 Scaleability + availability
• Partitions:
 Scaleability
• Packs
 Partition availability via fail-over
Farm
Clone Partition
Pack
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
40
Unpredictable Growth
The TerraServer Story:
• We expected 5 M hits per day
• We got 50 M hits on day 1
• We peak at 15-20 M hpd on a “hot” day
• Average 5 M hpd after 1 year
Most of us cannot predict demand
• Must be able to deal with NO demand
• Must be able to deal with HUGE demand
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
41
An Architecture for Internet Services?
Need to be able to add capacity
• New processing
• New storage
• New networking
Need continuous service
• Online change of all components (hardware and software)
• Multiple service sites
• Multiple network providers
Need great development tools
• Change the application several times per year.
• Add new services several times per year.
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
42
Premise: Each Site is a
Farm
Buy computing by the slice (brick):
• Rack of servers + disks.
Grow by adding slices
• Spread data and
computation
to new slices
Two growth styles:
• Clones: anonymous servers
• Parts+Packs: Partitions fail over within a pack
In both cases,
remote farm for disaster recovery
Switched
Ethernet
Switched
Ethernet
www.microsoft.com
(3)
search.microsoft.com
(1)
premium.microsoft.com
(1)
European Data Center
FTP
Download Server
(1)
SQL SERVERS
(2)
Router
msid.msn.com
(1)
MOSWest
Admin LAN
SQLNet
Feeder LAN
FDDI Ring
(MIS4)
Router
www.microsoft.com
(5)
Building 11
Live SQL Server
Router
home.microsoft.com
(5)
FDDI Ring
(MIS2)
www.microsoft.com
(4)
activex.microsoft.com
(2)
search.microsoft.com
(3)
register.microsoft.com
(2)
msid.msn.com
(1)
FDDI Ring
(MIS3)
www.microsoft.com
(3)
premium.microsoft.com
(1)
msid.msn.com
(1)
FDDI Ring
(MIS1)
www.microsoft.com
(4)
premium.microsoft.com
(2)
register.microsoft.com
(2)
msid.msn.com
(1) Primary
Gigaswitch
Secondary
Gigaswitch
Staging Servers
(7)
search.microsoft.com
(3)
support.microsoft.com
(2)
register.msn.com
(2)
1997 Microsoft.Com Farm
MOSWest
DMZ Staging Servers
Live SQL Servers
SQL Consolidators
Japan Data Center
www.microsoft.com
(3)
premium.microsoft.com
(1)
HTTP
Download Servers
(2) Router
search.microsoft.com
(2)
SQL SERVERS
(2)
msid.msn.com
(1)
FTP
Download Server
(1)
Router
Router
Router
Router
Router
Router
Router
Router
Internal WWW
SQL Reporting
home.microsoft.com
(4)
home.microsoft.com
(3)
home.microsoft.com
(2)
register.microsoft.com
(1)
support.microsoft.com
(1)
Internet
13
DS3
(45 Mb/Sec Each)
2
OC3
(45Mb/Sec Each)
2
Ethernet
(100 Mb/Sec Each)
cdm.microsoft.com
(1)
FTP Servers
Download
Replication
FTP.microsoft.com
(3)
0
Switched
Ethernet
Switched
Ethernet
www.microsoft.com
(3)
Switched
Ethernet
Switched
Ethernet
www.microsoft.com
(3)
search.microsoft.com
(1)
premium.microsoft.com
(1)
European Data Center
FTP
Download Server
(1)
SQL SERVERS
(2)
Router
msid.msn.com
(1)
MOSWest
Admin LAN
SQLNet
Feeder LAN
FDDI Ring
(MIS4)
Router
www.microsoft.com
(5)
Building 11
Live SQL Server
Router
home.microsoft.com
(5)
FDDI Ring
(MIS2)
www.microsoft.com
(4)
activex.microsoft.com
(2)
search.microsoft.com
(3)
register.microsoft.com
(2)
msid.msn.com
(1)
search.microsoft.com
(3)
register.microsoft.com
(2)
msid.msn.com
(1)
FDDI Ring
(MIS3)
www.microsoft.com
(3)
premium.microsoft.com
(1)
msid.msn.com
(1)
FDDI Ring
(MIS1)
www.microsoft.com
(4)
premium.microsoft.com
(2)
FDDI Ring
(MIS3)
www.microsoft.com
(3)
premium.microsoft.com
(1)
msid.msn.com
(1)
FDDI Ring
(MIS1)
www.microsoft.com
(4)
premium.microsoft.com
(2)
register.microsoft.com
(2)
msid.msn.com
(1) Primary
Gigaswitch
Secondary
Gigaswitch
register.microsoft.com
(2)
msid.msn.com
(1) Primary
Gigaswitch
Secondary
Gigaswitch
Staging Servers
(7)
search.microsoft.com
(3)
support.microsoft.com
(2)
register.msn.com
(2)
1997 Microsoft.Com Farm
MOSWest
DMZ Staging Servers
Live SQL Servers
SQL Consolidators
Live SQL Servers
SQL Consolidators
Japan Data Center
www.microsoft.com
(3)
premium.microsoft.com
(1)
HTTP
Download Servers
(2) Router
search.microsoft.com
(2)
SQL SERVERS
(2)
msid.msn.com
(1)
premium.microsoft.com
(1)
HTTP
Download Servers
(2) Router
search.microsoft.com
(2)
SQL SERVERS
(2)
msid.msn.com
(1)
FTP
Download Server
(1)
Router
Router
Router
Router
Router
Router
Router
Router
FTP
Download Server
(1)
Router
Router
Router
Router
Router
Router
Router
Router
Internal WWW
SQL Reporting
Internal WWW
SQL Reporting
home.microsoft.com
(4)
home.microsoft.com
(3)
home.microsoft.com
(2)
register.microsoft.com
(1)
support.microsoft.com
(1)
home.microsoft.com
(4)
home.microsoft.com
(3)
home.microsoft.com
(2)
register.microsoft.com
(1)
support.microsoft.com
(1)
Internet
13
DS3
(45 Mb/Sec Each)
2
OC3
(45Mb/Sec Each)
2
Ethernet
(100 Mb/Sec Each)
cdm.microsoft.com
(1)
FTP Servers
Download
Replication
FTP.microsoft.com
(3)
0
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
43
Scaleable Systems
Scale UP and Scale OUT
Everyone does both.
Choice is
• Size of a brick
• Clones or partitions
• Size of a pack
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
44
Everyone scales out
What’s the Brick?
1M$/slice
• IBM S390?
• Sun E 10,000?
 100 K$/slice
• Wintel 8X
 10 K$/slice
• Wintel 4x
1 K$/slice
• Wintel 1x
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
45
Clones: Availability+Scalability
Some applications are
• Read-mostly
• Low consistency requirements
• Modest storage requirement (less than 1TB)
Examples:
• HTTP web servers (IP sprayer/sieve + replication)
• LDAP servers (replication via gossip)
• App/compute servers or firewalls
Replicate app at all nodes (clones)
Spray requests across nodes.
Grow by adding clones
Fault tolerance: stop sending to dead
clone.
Growth: add a clone.
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
46
Facilities Clones Need
Automatic replication
• Applications (and system software)
• Data
Automatic request routing
• Spray or sieve
Management:
• Who is up?
• Update management & propagation
• Application monitoring.
Clones are very easy to manage:
• Rule of thumb: 100’s of clones per admin
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
47
Partitions for Scalability
Clones are not appropriate for some apps.
• Statefull apps do not replicate well
• high update rates do not replicate well
• Huge DBs (disk to expensive to clone)
Examples
• Email / chat / …
• Databases
Partition state among servers
Scalability (online):
• Partition split/merge
• Partitioning must be transparent to client.
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
48
Partitioned (aka. Clustered) Apps
Mail servers
• Perfectly partitionable
 Business Object Servers
• Partition by set of objects.
Parallel Databases
Transparent access to partitioned tables
Parallel Query
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
49
Packs for Availability
Each partition may fail (independent of others)
Partitions migrate to new node via fail-over
• Fail-over in seconds
Pack: the nodes supporting a partition
• VMS Cluster
• Tandem Process Pair
• SP2 HACMP
• Sysplex™
• WinNT MSCS (wolfpack)
Cluster In A Box
now commodity
Partitions grow in packs.
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
50
What Parts+Packs Need
Automatic partitioning (in dbms, mail, files,…)
• Location transparent
• Partition split/merge
• Grow without limits (100x10TB)
Simple failover model
• Partition migration is transparent
• MSCS-like model for services
Application-centric request routing
Management:
• Who is up?
• Automatic partition management (split/merge)
• Application monitoring.
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
51
Services on Clones & Partitions
Application provides a set of services
If cloned:
• Services are on subset of clones
If partitioned:
• Services run at each partition
System load balancing routes request to
• Any clone
• Correct partition.
• Routes around failures.
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
52
Farm pairs: Always Up
 Two farms
 Changes from one
sent to other
 When one farm fails
other provides service
 Masks
• Hardware/Software faults
• Operations tasks (reorganize, upgrade move
• Environmental faults (power fail)
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
53
Availabilty for a simple web site
Web Clients
Front End
Web File Store SQL Temp State
SQL Database
Packs for availability
Clones for availability
Load Balance
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
54
Cloned Packed file servers
Packed Partitions: Database Transparency
Farm Scale Out Scenarios
SQL Temp State
Web File StoreA
Cloned
Front Ends
(firewall, sprayer,
web server)
SQL Partition 3
The FARM: Clones and Packs of Partitions
Web
Clients
Web File StoreB
replication
SQL Database
SQL Partition 2 SQL Partition1
Load Balance
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
55
Clients
IIS Web Server
or other IP based services
Data Servers
SQL, Exchange, File
Network Load
Balancing
Clones
Cluster Service
Pack
COM+ Components
Component Load
Balancing (COM+)
Clones
Application Servers
Reliable, Scalable, Modular
1
2
32
3
…
1
2
8
…
4 3
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
56
Talk 2 (if there is time)
Terminology for scaleability
Farms of servers:
• Clones: identical
 Scaleability + availability
• Partitions:
 Scaleability
• Packs
 Partition availability via fail-over
Farm
Clone Partition
Pack
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
57
0
50
100
150
200
250
300
350
400
450
NT4
InProc
W2K
B1
inProc
W2K
RC1
InProc
W2K
RC2
InProc
1P 2P 4P 8P
0
50
100
150
200
250
300
350
400
450
NT4
InProc
W2K
B1
inProc
W2K
RC1
InProc
W2K
RC2
InProc
NT4
OOP
W2K
B1
OOP
W2K
RC1
OOP
W2K
RC2
OOP
1P 2P 4P 8P
Scalability: COM+ progress
serving 1,000-statement ASP’s (servelets)
Poor SMP Scaleability on IIS4 NT4
Big
improvements
from standard
Transaction
Processing
tricks
Out of Proc
(safe execution)
now much faster
than In Proc
was on IIS4
SPS: servelets per second
(ASPs served per second by IIS,
1,000 statement VBscript)
Shift from 4x200
Mhz to 8 450 Mhz
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
58
Scaleability:
So, What about the death of NT/Alpha?
Two simultaneous Compaq TPC-C numbers
 Intel Profusion
 NT/SQL/COM+
 550 Mhz
 8 Processors
 4 GB memory
 40,368 TPM-C @
18.46$/tpmC
 745 K$ 5-year cost
 Avail: 12/31/99
 Alpha
 Unix/Sybase/Tuexdo
 700 Mhz
 8 Processors
 16 GB memory
 42,437 TPM-C @
55.45 $/tpmC
 $2.35 M$ 5-year cost
 Avail: 10/18/99
200% more expense for 5% more performance?
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
59
Outline
Summary of what you heard. (10 min)
The database scene in general. (10 min)
Scaleability: Farms, Clones,Parts & Packs (15 min)
Microsoft DB research focus. (15 min)
• TerraServer (design and ops).
• RAGS.
• Data Mining
Q&A (10 min)
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
60
The TerraServer
http://www.terraserver.microsoft.com/
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
61
 Coverage: Range from 70ºN to 70ºS
today: 35% U.S., 1% outside U.S.
 Source Imagery:
• 4 TB 1sq meter/pixel Aerial (USGS - 60,000
46Mb B&W- 151Mb Color IR files)
• 1 TB 1.56 meter/pixel Satellite
(Spin-2 - 2400 300 Mb B&W)
 Display Imagery: 200x200 pixel images,
subsample to build image pyramid
 Store 5x compressed data
 Nav Tools:
• 1.5 m place names
• “Click-on” Coverage map
• Expedia & Virtual Globe
• Pick of the week
1.6x 1.6 km “city view”
.8 x .8 km 8m thumbnail
.4 x.4 km browse
200x200 m tile
Concept: User
navigates an ‘almost
seamless’ image of
earth
Database & application UI
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
62
browser
HTML
Java
Viewer
The Internet
Web Client
Image Delivery
Application
SQL Server
7
Microsoft
Site Server EE
Internet Information
Server 4.0
Image Provider Site(s)
TerraServer DB
Terra-Server
Stored Procedures
Internet
Information
Server 5.0
Image Server
Active Server Pages
(ADO)
MTS
TerraServer Web Site
Software:
Classic 3 Tier Design
SQL Server 7
24
20 (8/12)
46
46
Fire wall
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
63
SourceMeta
Image
Imagery
Image
Type
Search
Image
Search
Logical Schema
Scale
Job
Load
Job
Load
Mgmt
External
Link
External
Group
External
Geo
Famous
Category
Famous
Place
TerraServer
Terra
Database
TerraAdmin
Admin
Gazetteer
Country
Name
State
Name
Place
Name
Place
Type
Small
Place Name
Pyramid
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
64
TerraServer File Group Layout
Convert 324 disks to 28 RAID5 sets
plus 28 spare drives
Make 4 NT volumes (RAID 50)
595 GB per volume
Build 30 20GB files on each volume
DB is File Group of 120 files
E: F: G: H:
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
65
Hardware
100 Mbps
Ethernet Switch
DS3
Site
Servers
Internet
Map
Server
SPIN-2
Web Servers
2.9 TB Database Server
AlphaServer 8400 8x400.
10 GB RAM
324 StorageWorks disks
10 drive tape library
(STC Timber Wolf DLT7000 )
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
66
Backup and Recovery
• STK 9710 Tape robot
• SQL Server Backup &
Restore +
• Legato Networker
• Fast, incremental,
differential, online
• Clocked at 80 MBps (peak)
(~ 200 GB/hr)
Restore
• Fast, incremental (file
oriented), not online.
Load & Backup&Recovery
Performance
Data Bytes Backed Up 1.2 TB
Total Time 7.25 Hours
Number of Tapes Consumed 27
Total Tape Drives 10
Data ThroughPut 168 GB/Ho
Average ThroughPut Per Device 16.8 GB/Hour
Average Throughput Per Device 4.97 MB/Sec
NTFS Logical Volumes 2
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
67
BAD OLD Load
DLT
Tape
“tar”
Drop’N’ DoJob
Wait 4
Load
LoadMgr
DB
100mbit
EtherSwitch
108
9.1 GB
Drives
Enterprise Storage Array
Alpha
Server
8400
108
9.1 GB
Drives
108
9.1 GB
Drives
STC
DLT
Tape
Library
60
4.3 GB
Drives
Alpha
Server
4100
ESA
Alpha
Server
4100
LoadMgr
DLT
Tape
NT
Backup
ImgCutter
Drop’N’
Images
10: ImgCutter
20: Partition
30: ThumbImg
40: BrowseImg
45: JumpImg
50: TileImg
55: Meta Data
60: Tile Meta
70: Img Meta
80: Update Place
...
LoadMgr
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
68
New Image Load and Update
ODBC Tx
TerraLoader
ODBC TX
TerraServer
SQL
DBMS
DLT
Tape
“tar”
Metadata
Load DB
Active Server Pages
Cut & Load
Scheduling
System
Image
Cutter
Merge
ODBC Tx
Dither
Image Pyramid
From base
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
70
TerraServer Daily Traffic
Jun 22, 1998 thru June 22, 1999
0
10M
20M
30M
Count
Sessions
Hit
Page View
DB Query
Image
After a Year:
1 TB of data
750 M records
2.3 billion Hits
2.0 billion DB Queries
1.7 billion Images sent
368 million Page Views
99.93% DB Availability
3rd design now Online
Built and operated by
team of 4 people
TotalTime (Hours)
0
720
1440
2160
2880
3600
4320
5040
5760
6480
7200
7920
8640
Up
Down Time
(Hours:minutes)
0:00
0:30
1:00
1:30
2:00
2:30
3:00
3:30
4:00
4:30
5:00
5:30
6:00
Scheduled
HW+Software
Operations
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
74
What TerraServer Shows
Can serve huge databases on Internet
for about a penny a page view
mostly phone bill (!).
Advertising pays more than a penny a page.
Commodity tools do scale fairly far.
A few people (3 developers, 1 operator)
using power tools
can build an impressive web site
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
75
Outline
Summary of what you heard. (20 min)
The database scene in general. (10 min)
Scaleability: Packs & Mobs (10 min)
Microsoft DB research focus. (15 min)
• TerraServer (design and ops).
• RAGS.
• Data Mining
Q&A (15 min)
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
76
Automatic Testing
60% of Microsoft R&D is testing.
What can research do to help?
• beyond joining the 500,000 Win2K beta testers
Test generation robot:
• Make up SQL queries
• Send them to SQL Server,
Oracle, DB2, Informix,…
• If answer is the same, great,
if not there is a problem
Also good for stress tests
Found MANY bugs in our products (all fixed).
Found MANY bugs in other’s products.
Very valuable tool.
 MSR-TR-98-21 Massive Stochastic Testing of SQL, Slutz, Don
http://research.microsoft.com/scripts/pubDB/pubsasp.asp?RecordID=175
W X Y Z
1672 1672 1672 1672
232 234 241 31
1 1 1 1
31 15 12 28
1 12 5 116
0 29 32 4
18 18 19 25
45 19 18 113
Error
All four
agree 84%
Problem with
intermediate
table.
Case
W,X, and
agree 95%
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
77
Some Tera-Byte Databases
Kilo
Mega
Giga
Tera
Peta
Exa
Zetta
Yotta
The Web: 1 TB of HTML
TerraServer 1 TB of images
Several other 1 TB (file) servers
Hotmail: 20 TB of email
Sloan Digital Sky Survey:
40 TB raw, 2 TB cooked
EOS/DIS (picture of planet each week)
• 15 PB by 2007
Federal Clearing house: images of checks
• 15 PB by 2006 (7 year history)
Nuclear Stockpile Stewardship Program
• 10 Exabytes (???!!)
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
78
Data Mining
Find interesting structure (patterns, relationships) in data
• Prediction
• Segmentation (clustering)
• Dependency modeling (find distribution)
• Summarization
• Trend and change detection and modeling
Allow user to state the query in terms of the business
logic
• User does not speak statistics or SQL
Use data to build predictors
• regression, classification, segmentation etc.
Generate summaries and reports for insight
• find “easy to describe” segments in data automatically
• find segments not known to analyst
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
79
Data Mining:
Microsoft SiteServer Commerce 3.0
Intelligent Cross-sell
Based on:
• Historical sales baskets in
stores
• Contents of current shopper
basket
• Browsing behavior of
shopper
Predict: ranking of products
in store likely to be most
interesting to shopper.
Http://www.holtoutlet.com/outlet4
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
80
Mail to 25% and capture 40%
400% improved response!
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
0.1%
0.2%
0.3%
0.6%
1.3%
5.3%
6.7%
25.5%
34.5%
43.8%
56.9%
68.5%
94.8%
98.5%
100.0%
% mailed
%
Captured
of
true
targets
Real data drawn from a Microsoft marketing example
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
81
How do people use
www.microsoft.com?
Data Mining
(Clustering)
Engine
User
browsing
data
100M hits per day
14M users/week X segments
Cluster
Visualizer
Wizard
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
82
Jim Gray, Research and Microsoft SQL Server
Microsoft Research http://research.Microsoft.com/~gray/talks/
3 October 1999
Chicago, Ill.
83
Outline
Summary of what you heard. (10 min)
The database scene in general. (10 min)
Scaleability: Farms, Clones,Parts & Packs (15 min)
Microsoft DB research focus. (15 min)
• TerraServer (design and ops).
• RAGS.
• Data Mining
Q&A (10 min)

More Related Content

Similar to Gray_Compass99.ppt

Simplifying Change Data Capture using Databricks Delta
Simplifying Change Data Capture using Databricks DeltaSimplifying Change Data Capture using Databricks Delta
Simplifying Change Data Capture using Databricks Delta
Databricks
 
Data Access Patterns
Data Access PatternsData Access Patterns
Data Access Patterns
Amazon Web Services
 
Off-Label Data Mesh: A Prescription for Healthier Data
Off-Label Data Mesh: A Prescription for Healthier DataOff-Label Data Mesh: A Prescription for Healthier Data
Off-Label Data Mesh: A Prescription for Healthier Data
HostedbyConfluent
 
Application Development and Data Modeling on Amazon DynamoDB
Application Development and Data Modeling on Amazon DynamoDBApplication Development and Data Modeling on Amazon DynamoDB
Application Development and Data Modeling on Amazon DynamoDB
Amazon Web Services Japan
 
Graph Database Use Cases - StampedeCon 2015
Graph Database Use Cases - StampedeCon 2015Graph Database Use Cases - StampedeCon 2015
Graph Database Use Cases - StampedeCon 2015
StampedeCon
 
Graph database Use Cases
Graph database Use CasesGraph database Use Cases
Graph database Use Cases
Max De Marzi
 
Sound cloud - User & Partner Conference - AT Internet
Sound cloud - User & Partner Conference - AT InternetSound cloud - User & Partner Conference - AT Internet
Sound cloud - User & Partner Conference - AT Internet
AT Internet
 
Apache IOTDB: a Time Series Database for Industrial IoT
Apache IOTDB: a Time Series Database for Industrial IoTApache IOTDB: a Time Series Database for Industrial IoT
Apache IOTDB: a Time Series Database for Industrial IoT
jixuan1989
 
What is a distributed data science pipeline. how with apache spark and friends.
What is a distributed data science pipeline. how with apache spark and friends.What is a distributed data science pipeline. how with apache spark and friends.
What is a distributed data science pipeline. how with apache spark and friends.
Andy Petrella
 
ISBG 2016 - XPages on IBM Bluemix
ISBG 2016 - XPages on IBM BluemixISBG 2016 - XPages on IBM Bluemix
ISBG 2016 - XPages on IBM Bluemix
Oliver Busse
 
Towards a rebirth of data science (by Data Fellas)
Towards a rebirth of data science (by Data Fellas)Towards a rebirth of data science (by Data Fellas)
Towards a rebirth of data science (by Data Fellas)
Andy Petrella
 
Large scale, interactive ad-hoc queries over different datastores with Apache...
Large scale, interactive ad-hoc queries over different datastores with Apache...Large scale, interactive ad-hoc queries over different datastores with Apache...
Large scale, interactive ad-hoc queries over different datastores with Apache...
jaxLondonConference
 
Database 101 on IBM i
Database 101 on IBM iDatabase 101 on IBM i
Database 101 on IBM i
HelpSystems
 
Understanding The Azure Platform November 09
Understanding The Azure Platform   November 09Understanding The Azure Platform   November 09
Understanding The Azure Platform November 09
DavidGristwood
 
ITCamp 2013 - Cristian Lefter - Transact-SQL from 0 to SQL Server 2012
ITCamp 2013 - Cristian Lefter - Transact-SQL from 0 to SQL Server 2012ITCamp 2013 - Cristian Lefter - Transact-SQL from 0 to SQL Server 2012
ITCamp 2013 - Cristian Lefter - Transact-SQL from 0 to SQL Server 2012
ITCamp
 
Introduction to Text Mining and Visualization with Interactive Web Application
Introduction to Text Mining and Visualization with Interactive Web ApplicationIntroduction to Text Mining and Visualization with Interactive Web Application
Introduction to Text Mining and Visualization with Interactive Web Application
Olga Scrivner
 
Big Data Analytics: Finding diamonds in the rough with Azure
Big Data Analytics: Finding diamonds in the rough with AzureBig Data Analytics: Finding diamonds in the rough with Azure
Big Data Analytics: Finding diamonds in the rough with Azure
Christos Charmatzis
 
Applied Enterprise Semantic Search 201305
Applied Enterprise Semantic Search 201305Applied Enterprise Semantic Search 201305
Applied Enterprise Semantic Search 201305
Mark Tabladillo
 
The Google Hacking Database: A Key Resource to Exposing Vulnerabilities
The Google Hacking Database: A Key Resource to Exposing VulnerabilitiesThe Google Hacking Database: A Key Resource to Exposing Vulnerabilities
The Google Hacking Database: A Key Resource to Exposing Vulnerabilities
TechWell
 
Synchronicity: Just-In-Time Discovery of Lost Web Pages
Synchronicity: Just-In-Time Discovery of Lost Web PagesSynchronicity: Just-In-Time Discovery of Lost Web Pages
Synchronicity: Just-In-Time Discovery of Lost Web Pages
Michael Nelson
 

Similar to Gray_Compass99.ppt (20)

Simplifying Change Data Capture using Databricks Delta
Simplifying Change Data Capture using Databricks DeltaSimplifying Change Data Capture using Databricks Delta
Simplifying Change Data Capture using Databricks Delta
 
Data Access Patterns
Data Access PatternsData Access Patterns
Data Access Patterns
 
Off-Label Data Mesh: A Prescription for Healthier Data
Off-Label Data Mesh: A Prescription for Healthier DataOff-Label Data Mesh: A Prescription for Healthier Data
Off-Label Data Mesh: A Prescription for Healthier Data
 
Application Development and Data Modeling on Amazon DynamoDB
Application Development and Data Modeling on Amazon DynamoDBApplication Development and Data Modeling on Amazon DynamoDB
Application Development and Data Modeling on Amazon DynamoDB
 
Graph Database Use Cases - StampedeCon 2015
Graph Database Use Cases - StampedeCon 2015Graph Database Use Cases - StampedeCon 2015
Graph Database Use Cases - StampedeCon 2015
 
Graph database Use Cases
Graph database Use CasesGraph database Use Cases
Graph database Use Cases
 
Sound cloud - User & Partner Conference - AT Internet
Sound cloud - User & Partner Conference - AT InternetSound cloud - User & Partner Conference - AT Internet
Sound cloud - User & Partner Conference - AT Internet
 
Apache IOTDB: a Time Series Database for Industrial IoT
Apache IOTDB: a Time Series Database for Industrial IoTApache IOTDB: a Time Series Database for Industrial IoT
Apache IOTDB: a Time Series Database for Industrial IoT
 
What is a distributed data science pipeline. how with apache spark and friends.
What is a distributed data science pipeline. how with apache spark and friends.What is a distributed data science pipeline. how with apache spark and friends.
What is a distributed data science pipeline. how with apache spark and friends.
 
ISBG 2016 - XPages on IBM Bluemix
ISBG 2016 - XPages on IBM BluemixISBG 2016 - XPages on IBM Bluemix
ISBG 2016 - XPages on IBM Bluemix
 
Towards a rebirth of data science (by Data Fellas)
Towards a rebirth of data science (by Data Fellas)Towards a rebirth of data science (by Data Fellas)
Towards a rebirth of data science (by Data Fellas)
 
Large scale, interactive ad-hoc queries over different datastores with Apache...
Large scale, interactive ad-hoc queries over different datastores with Apache...Large scale, interactive ad-hoc queries over different datastores with Apache...
Large scale, interactive ad-hoc queries over different datastores with Apache...
 
Database 101 on IBM i
Database 101 on IBM iDatabase 101 on IBM i
Database 101 on IBM i
 
Understanding The Azure Platform November 09
Understanding The Azure Platform   November 09Understanding The Azure Platform   November 09
Understanding The Azure Platform November 09
 
ITCamp 2013 - Cristian Lefter - Transact-SQL from 0 to SQL Server 2012
ITCamp 2013 - Cristian Lefter - Transact-SQL from 0 to SQL Server 2012ITCamp 2013 - Cristian Lefter - Transact-SQL from 0 to SQL Server 2012
ITCamp 2013 - Cristian Lefter - Transact-SQL from 0 to SQL Server 2012
 
Introduction to Text Mining and Visualization with Interactive Web Application
Introduction to Text Mining and Visualization with Interactive Web ApplicationIntroduction to Text Mining and Visualization with Interactive Web Application
Introduction to Text Mining and Visualization with Interactive Web Application
 
Big Data Analytics: Finding diamonds in the rough with Azure
Big Data Analytics: Finding diamonds in the rough with AzureBig Data Analytics: Finding diamonds in the rough with Azure
Big Data Analytics: Finding diamonds in the rough with Azure
 
Applied Enterprise Semantic Search 201305
Applied Enterprise Semantic Search 201305Applied Enterprise Semantic Search 201305
Applied Enterprise Semantic Search 201305
 
The Google Hacking Database: A Key Resource to Exposing Vulnerabilities
The Google Hacking Database: A Key Resource to Exposing VulnerabilitiesThe Google Hacking Database: A Key Resource to Exposing Vulnerabilities
The Google Hacking Database: A Key Resource to Exposing Vulnerabilities
 
Synchronicity: Just-In-Time Discovery of Lost Web Pages
Synchronicity: Just-In-Time Discovery of Lost Web PagesSynchronicity: Just-In-Time Discovery of Lost Web Pages
Synchronicity: Just-In-Time Discovery of Lost Web Pages
 

Recently uploaded

Let's Integrate MuleSoft RPA, COMPOSER, APM with AWS IDP along with Slack
Let's Integrate MuleSoft RPA, COMPOSER, APM with AWS IDP along with SlackLet's Integrate MuleSoft RPA, COMPOSER, APM with AWS IDP along with Slack
Let's Integrate MuleSoft RPA, COMPOSER, APM with AWS IDP along with Slack
shyamraj55
 
Monitoring and Managing Anomaly Detection on OpenShift.pdf
Monitoring and Managing Anomaly Detection on OpenShift.pdfMonitoring and Managing Anomaly Detection on OpenShift.pdf
Monitoring and Managing Anomaly Detection on OpenShift.pdf
Tosin Akinosho
 
Trusted Execution Environment for Decentralized Process Mining
Trusted Execution Environment for Decentralized Process MiningTrusted Execution Environment for Decentralized Process Mining
Trusted Execution Environment for Decentralized Process Mining
LucaBarbaro3
 
A Comprehensive Guide to DeFi Development Services in 2024
A Comprehensive Guide to DeFi Development Services in 2024A Comprehensive Guide to DeFi Development Services in 2024
A Comprehensive Guide to DeFi Development Services in 2024
Intelisync
 
UI5 Controls simplified - UI5con2024 presentation
UI5 Controls simplified - UI5con2024 presentationUI5 Controls simplified - UI5con2024 presentation
UI5 Controls simplified - UI5con2024 presentation
Wouter Lemaire
 
Recommendation System using RAG Architecture
Recommendation System using RAG ArchitectureRecommendation System using RAG Architecture
Recommendation System using RAG Architecture
fredae14
 
Driving Business Innovation: Latest Generative AI Advancements & Success Story
Driving Business Innovation: Latest Generative AI Advancements & Success StoryDriving Business Innovation: Latest Generative AI Advancements & Success Story
Driving Business Innovation: Latest Generative AI Advancements & Success Story
Safe Software
 
Unlock the Future of Search with MongoDB Atlas_ Vector Search Unleashed.pdf
Unlock the Future of Search with MongoDB Atlas_ Vector Search Unleashed.pdfUnlock the Future of Search with MongoDB Atlas_ Vector Search Unleashed.pdf
Unlock the Future of Search with MongoDB Atlas_ Vector Search Unleashed.pdf
Malak Abu Hammad
 
June Patch Tuesday
June Patch TuesdayJune Patch Tuesday
June Patch Tuesday
Ivanti
 
WeTestAthens: Postman's AI & Automation Techniques
WeTestAthens: Postman's AI & Automation TechniquesWeTestAthens: Postman's AI & Automation Techniques
WeTestAthens: Postman's AI & Automation Techniques
Postman
 
HCL Notes und Domino Lizenzkostenreduzierung in der Welt von DLAU
HCL Notes und Domino Lizenzkostenreduzierung in der Welt von DLAUHCL Notes und Domino Lizenzkostenreduzierung in der Welt von DLAU
HCL Notes und Domino Lizenzkostenreduzierung in der Welt von DLAU
panagenda
 
Azure API Management to expose backend services securely
Azure API Management to expose backend services securelyAzure API Management to expose backend services securely
Azure API Management to expose backend services securely
Dinusha Kumarasiri
 
Letter and Document Automation for Bonterra Impact Management (fka Social Sol...
Letter and Document Automation for Bonterra Impact Management (fka Social Sol...Letter and Document Automation for Bonterra Impact Management (fka Social Sol...
Letter and Document Automation for Bonterra Impact Management (fka Social Sol...
Jeffrey Haguewood
 
TrustArc Webinar - 2024 Global Privacy Survey
TrustArc Webinar - 2024 Global Privacy SurveyTrustArc Webinar - 2024 Global Privacy Survey
TrustArc Webinar - 2024 Global Privacy Survey
TrustArc
 
Presentation of the OECD Artificial Intelligence Review of Germany
Presentation of the OECD Artificial Intelligence Review of GermanyPresentation of the OECD Artificial Intelligence Review of Germany
Presentation of the OECD Artificial Intelligence Review of Germany
innovationoecd
 
HCL Notes and Domino License Cost Reduction in the World of DLAU
HCL Notes and Domino License Cost Reduction in the World of DLAUHCL Notes and Domino License Cost Reduction in the World of DLAU
HCL Notes and Domino License Cost Reduction in the World of DLAU
panagenda
 
Taking AI to the Next Level in Manufacturing.pdf
Taking AI to the Next Level in Manufacturing.pdfTaking AI to the Next Level in Manufacturing.pdf
Taking AI to the Next Level in Manufacturing.pdf
ssuserfac0301
 
How to Interpret Trends in the Kalyan Rajdhani Mix Chart.pdf
How to Interpret Trends in the Kalyan Rajdhani Mix Chart.pdfHow to Interpret Trends in the Kalyan Rajdhani Mix Chart.pdf
How to Interpret Trends in the Kalyan Rajdhani Mix Chart.pdf
Chart Kalyan
 
Skybuffer SAM4U tool for SAP license adoption
Skybuffer SAM4U tool for SAP license adoptionSkybuffer SAM4U tool for SAP license adoption
Skybuffer SAM4U tool for SAP license adoption
Tatiana Kojar
 
Deep Dive: AI-Powered Marketing to Get More Leads and Customers with HyperGro...
Deep Dive: AI-Powered Marketing to Get More Leads and Customers with HyperGro...Deep Dive: AI-Powered Marketing to Get More Leads and Customers with HyperGro...
Deep Dive: AI-Powered Marketing to Get More Leads and Customers with HyperGro...
saastr
 

Recently uploaded (20)

Let's Integrate MuleSoft RPA, COMPOSER, APM with AWS IDP along with Slack
Let's Integrate MuleSoft RPA, COMPOSER, APM with AWS IDP along with SlackLet's Integrate MuleSoft RPA, COMPOSER, APM with AWS IDP along with Slack
Let's Integrate MuleSoft RPA, COMPOSER, APM with AWS IDP along with Slack
 
Monitoring and Managing Anomaly Detection on OpenShift.pdf
Monitoring and Managing Anomaly Detection on OpenShift.pdfMonitoring and Managing Anomaly Detection on OpenShift.pdf
Monitoring and Managing Anomaly Detection on OpenShift.pdf
 
Trusted Execution Environment for Decentralized Process Mining
Trusted Execution Environment for Decentralized Process MiningTrusted Execution Environment for Decentralized Process Mining
Trusted Execution Environment for Decentralized Process Mining
 
A Comprehensive Guide to DeFi Development Services in 2024
A Comprehensive Guide to DeFi Development Services in 2024A Comprehensive Guide to DeFi Development Services in 2024
A Comprehensive Guide to DeFi Development Services in 2024
 
UI5 Controls simplified - UI5con2024 presentation
UI5 Controls simplified - UI5con2024 presentationUI5 Controls simplified - UI5con2024 presentation
UI5 Controls simplified - UI5con2024 presentation
 
Recommendation System using RAG Architecture
Recommendation System using RAG ArchitectureRecommendation System using RAG Architecture
Recommendation System using RAG Architecture
 
Driving Business Innovation: Latest Generative AI Advancements & Success Story
Driving Business Innovation: Latest Generative AI Advancements & Success StoryDriving Business Innovation: Latest Generative AI Advancements & Success Story
Driving Business Innovation: Latest Generative AI Advancements & Success Story
 
Unlock the Future of Search with MongoDB Atlas_ Vector Search Unleashed.pdf
Unlock the Future of Search with MongoDB Atlas_ Vector Search Unleashed.pdfUnlock the Future of Search with MongoDB Atlas_ Vector Search Unleashed.pdf
Unlock the Future of Search with MongoDB Atlas_ Vector Search Unleashed.pdf
 
June Patch Tuesday
June Patch TuesdayJune Patch Tuesday
June Patch Tuesday
 
WeTestAthens: Postman's AI & Automation Techniques
WeTestAthens: Postman's AI & Automation TechniquesWeTestAthens: Postman's AI & Automation Techniques
WeTestAthens: Postman's AI & Automation Techniques
 
HCL Notes und Domino Lizenzkostenreduzierung in der Welt von DLAU
HCL Notes und Domino Lizenzkostenreduzierung in der Welt von DLAUHCL Notes und Domino Lizenzkostenreduzierung in der Welt von DLAU
HCL Notes und Domino Lizenzkostenreduzierung in der Welt von DLAU
 
Azure API Management to expose backend services securely
Azure API Management to expose backend services securelyAzure API Management to expose backend services securely
Azure API Management to expose backend services securely
 
Letter and Document Automation for Bonterra Impact Management (fka Social Sol...
Letter and Document Automation for Bonterra Impact Management (fka Social Sol...Letter and Document Automation for Bonterra Impact Management (fka Social Sol...
Letter and Document Automation for Bonterra Impact Management (fka Social Sol...
 
TrustArc Webinar - 2024 Global Privacy Survey
TrustArc Webinar - 2024 Global Privacy SurveyTrustArc Webinar - 2024 Global Privacy Survey
TrustArc Webinar - 2024 Global Privacy Survey
 
Presentation of the OECD Artificial Intelligence Review of Germany
Presentation of the OECD Artificial Intelligence Review of GermanyPresentation of the OECD Artificial Intelligence Review of Germany
Presentation of the OECD Artificial Intelligence Review of Germany
 
HCL Notes and Domino License Cost Reduction in the World of DLAU
HCL Notes and Domino License Cost Reduction in the World of DLAUHCL Notes and Domino License Cost Reduction in the World of DLAU
HCL Notes and Domino License Cost Reduction in the World of DLAU
 
Taking AI to the Next Level in Manufacturing.pdf
Taking AI to the Next Level in Manufacturing.pdfTaking AI to the Next Level in Manufacturing.pdf
Taking AI to the Next Level in Manufacturing.pdf
 
How to Interpret Trends in the Kalyan Rajdhani Mix Chart.pdf
How to Interpret Trends in the Kalyan Rajdhani Mix Chart.pdfHow to Interpret Trends in the Kalyan Rajdhani Mix Chart.pdf
How to Interpret Trends in the Kalyan Rajdhani Mix Chart.pdf
 
Skybuffer SAM4U tool for SAP license adoption
Skybuffer SAM4U tool for SAP license adoptionSkybuffer SAM4U tool for SAP license adoption
Skybuffer SAM4U tool for SAP license adoption
 
Deep Dive: AI-Powered Marketing to Get More Leads and Customers with HyperGro...
Deep Dive: AI-Powered Marketing to Get More Leads and Customers with HyperGro...Deep Dive: AI-Powered Marketing to Get More Leads and Customers with HyperGro...
Deep Dive: AI-Powered Marketing to Get More Leads and Customers with HyperGro...
 

Gray_Compass99.ppt

  • 1. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 1 Microsoft SQL Server, Scalability, & Database Research Jim Gray Researcher Microsoft Corporation
  • 2. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 2 Outline Summary of what you heard. (10 min) The database scene in general. (10 min) Scaleability: Farms, Clones,Parts & Packs (15 min) Microsoft DB research focus. (15 min) • TerraServer (design and ops). • RAGS. • Data Mining Q&A (10 min)
  • 3. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 3 Organizations Are Going Online Building a digital nervous system. Inexpensive hardware means huge databases are possible. But, we are drowning in data. Databases help organize information. Microsoft’s goal: • Information at your fingertips. • Make it easy to capture, manage, and analyze information.
  • 4. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 4 Microsoft SQL Server 7 Goals Easy  Dynamic self management  Multi-site management  Operation Scripting  Job scheduling and execution  Alert/response management  Scriptable Install+upgrade  DBA profiling/tuning tools  Unicode  English Language Query  Integrated with NT Security  Integrated with NT files Scalability Data Warehousing
  • 5. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 11 Scalability  Win9x/NTW version  Dynamic row-level locking  Improved query optimizer  Intra-query parallelism  VLDB improvements  Replication improvements  Distributed query  High Availability Clusters Easy Scalability Data Warehousing
  • 6. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 12 Scale Down to Windows 95-98 Full function (same as NTW) Integration with Access 97 MSDE in Office2000 and MSDN WinCE version demonstrated
  • 7. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 13 Replication Transactional and Merge Remote update ODBC and OLE DB subscribers Wizards Performance 2PC, RPC Subscriber DB2 CICS Subscriber Subscriber VSAM OS 390 DB2 Publisher Updating Subscriber (immediate updates) Distributor Subscriber
  • 8. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 14 Query Processor Enhancements Parallelism Improved scan, fetch, & sort Smart hash & merge join Large joins & grouping Better query optimization Multi-index operations Automatic statistics maintenance Distributed Query Heterogeneous Query Focus on Complex Queries
  • 9. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 15 •# of emp. per group •total inc. per group Local Agg. 4 x 50 rows + + + + Disks 50,000 rows Global Agg. Result 50 rows + Parallel Query SMP & Disk Parallelism Plus Distributed Plus Hash Join (fanciest on the planet) Plus Optimized Partitioned views
  • 10. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 16 Distributed Heterogeneous Queries Data Fusion / Integration Join spread sheets, databases, directories, Text DBs etc. Any source that exposes OLE DB interfaces SQL Server as gateway, even on the desktop Database (DB2, VSAM, Oracle, …) Spreadsheet Photos Mail Maps Documents and the Web Directory Service SQL 7.0 Query Processor
  • 11. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 17 Utilities The Key to LARGE Databases  Backup • Fuzzy • Parallel • Incremental • Restartable Recovery • Fast • File granularity Reorganize • shrinks file • reclusters file Auto-Repair Index creation ~2x faster than 6.5 DBCC • not required, • a good practice • 5x - 100x faster 0 10 20 30 40 50 60 1 2 3 4 SQL Server 6.5 SQL Server 7.0 # of indices Recovery time ( secs )
  • 12. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 20 Data Warehousing  Warehousing Framework  Visual data modeler  Microsoft repository  Data transformation services (DTS)  Plato & Dcube - Multi Dimensional Data Cubes  English query 2.0 Easy Scalability Data Warehousing
  • 13. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 22 Data Warehouse / Data Analysis Data Transformation Services to get data into the warehouse CUBE (OLE/DB OLAP) to analyze data Operational Data Extact & Load Data Warehouse Storage OLAP
  • 14. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 25 Source table Partition 1 ROLAP Partition 2 Partition 3 ROLAP Europe USA Asia MD SQL SQL Plato and Data Cube and HOLAP “Plato” server “Plato” Designer Dcube Client app User 1 Dcube Client app User 2 RED WHITE BLUE By Color By Make & Year By Color & Year By Make By Year Sum
  • 15. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 26 English Query
  • 16. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 27 Easy Scalable Data Warehousing
  • 17. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. “Shiloh” The Next SQL Server Shiloh (H1’00) - Strengthen Position • Data Warehousing leadership  Materialized Views  Cascading Referential Integrity (#1 requested user-group feature)  XML support • Scalability  WinCE support  W2K VLM (36 and 64 bit)  Multi-instance support Yukon – Next Big Step • Scalability (Clusters, Partitions) • Programmability • Ease of Use (Self Tuning, Auto Config)
  • 18. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 29 Outline Summary of what you heard. (10 min) The database scene in general. (10 min) Scaleability: Farms, Clones,Parts & Packs (15 min) Microsoft DB research focus. (15 min) • TerraServer (design and ops). • RAGS. • Data Mining Q&A (10 min)
  • 19. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 30 Info Capture You can record everything you see or hear or read. What will you do with it? How will you organize & analyze it? Most data will never be seen Analysis an summarization are key technologies Video 8 PB per lifetime (10GBph) Audio 30 TB (10KBps) Read or write: 8 GB (words) See: http://www.lesk.com/mlesk/ksg97/ksg.html Yotta Zetta Exa Peta Tera Giga Mega Kilo A Book .Movi e All books (words) All Books MultiMedia Everything ! Recorded A Photo
  • 20. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 31 Data Tidal Wave  Seagate 47GB drive @ 783$ (= 1.7 ¢/mb) • 100 GB penny per MB drive coming in 2000  10 $/GB = 10 k$/ Terabyte! • “Everyone” can afford one  What’s a terror bite? • If you sell ten billion items a year (e.g Wal-Mart) • And you record 100 bytes on each one • Then you get a TeraByte/year  Where will the terror bytes come from? • Multimedia (like the TerraServer) and...
  • 21. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 32 Reducing Data’s Cost-of-Ownership Self-Managing data Cost of ownership: One admin/TB (100K$ vs 10K$) Admin cost exceeds storage cost. SQL 7: Suggests indices Migrates data away from end of file Truncates file Someday: Automatic move files to balance disks Online defragmentation & restructuring Online physical redesign
  • 22. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 33 OBJECT RELATIONAL The Next Great DBMS Wave All DB vendors have added objects to DB Microsoft is adding DBs to Objects Integration with COM+ Gives user-defined types and objects Plug-ins will be Billion dollar industry • Blades for SQL Server razor
  • 23. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 34 Why Is XML Important? Self-describing data “ABC47-Z”, “100”, “STL”, “C”, “3”, “28” Data stream in a typical interface… <INVENTORY> <PART_NUM>ABC47-Z</PART_NUM> <QUANTITY>100</QUANTITY> <WAREHOUSE>STL</WAREHOUSE> <ZONE>C</ZONE> <AISLE>3</AISLE> <BIN>28</BIN> </INVENTORY> Same data stream in XML…
  • 24. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 35 table.xsl bar.xsl art.xsl
  • 25. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 36 XML Applications Exposing Software as a “Service” • Websites without UI’s • Exposed services with common scheme • Integration points at the enterprise, value- chain, workgroup, desktop and intelligent gizmo “levels” B2B value chains • Uses XML to transmit wide range of date to a broad set of stakeholders (regulatory agencies, suppliers, customers, etc.). • Leverage for prior efforts like EDI • BizTalk a key industry effort in this regard
  • 26. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 37 XML Message XML Document XML Message XML: BizTalk Framework Order Processing MVS CICS SAP R/3 XML XML Service Interface Browser Client Apps New Form Factors XML Document Another Service JD Edwards XML XML Library www.biztalk.org XML XML schema
  • 27. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 38 Outline Summary of what you heard. (20 min) The database scene in general. (10 min) Scaleability: Farms, Clones,Parts & Packs (10 min) Microsoft DB research focus. (15 min) • TerraServer (design and ops). • RAGS. • Data Mining Q&A (15 min)
  • 28. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 39 Terminology for scaleability Farms of servers: • Clones: identical  Scaleability + availability • Partitions:  Scaleability • Packs  Partition availability via fail-over Farm Clone Partition Pack
  • 29. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 40 Unpredictable Growth The TerraServer Story: • We expected 5 M hits per day • We got 50 M hits on day 1 • We peak at 15-20 M hpd on a “hot” day • Average 5 M hpd after 1 year Most of us cannot predict demand • Must be able to deal with NO demand • Must be able to deal with HUGE demand
  • 30. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 41 An Architecture for Internet Services? Need to be able to add capacity • New processing • New storage • New networking Need continuous service • Online change of all components (hardware and software) • Multiple service sites • Multiple network providers Need great development tools • Change the application several times per year. • Add new services several times per year.
  • 31. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 42 Premise: Each Site is a Farm Buy computing by the slice (brick): • Rack of servers + disks. Grow by adding slices • Spread data and computation to new slices Two growth styles: • Clones: anonymous servers • Parts+Packs: Partitions fail over within a pack In both cases, remote farm for disaster recovery Switched Ethernet Switched Ethernet www.microsoft.com (3) search.microsoft.com (1) premium.microsoft.com (1) European Data Center FTP Download Server (1) SQL SERVERS (2) Router msid.msn.com (1) MOSWest Admin LAN SQLNet Feeder LAN FDDI Ring (MIS4) Router www.microsoft.com (5) Building 11 Live SQL Server Router home.microsoft.com (5) FDDI Ring (MIS2) www.microsoft.com (4) activex.microsoft.com (2) search.microsoft.com (3) register.microsoft.com (2) msid.msn.com (1) FDDI Ring (MIS3) www.microsoft.com (3) premium.microsoft.com (1) msid.msn.com (1) FDDI Ring (MIS1) www.microsoft.com (4) premium.microsoft.com (2) register.microsoft.com (2) msid.msn.com (1) Primary Gigaswitch Secondary Gigaswitch Staging Servers (7) search.microsoft.com (3) support.microsoft.com (2) register.msn.com (2) 1997 Microsoft.Com Farm MOSWest DMZ Staging Servers Live SQL Servers SQL Consolidators Japan Data Center www.microsoft.com (3) premium.microsoft.com (1) HTTP Download Servers (2) Router search.microsoft.com (2) SQL SERVERS (2) msid.msn.com (1) FTP Download Server (1) Router Router Router Router Router Router Router Router Internal WWW SQL Reporting home.microsoft.com (4) home.microsoft.com (3) home.microsoft.com (2) register.microsoft.com (1) support.microsoft.com (1) Internet 13 DS3 (45 Mb/Sec Each) 2 OC3 (45Mb/Sec Each) 2 Ethernet (100 Mb/Sec Each) cdm.microsoft.com (1) FTP Servers Download Replication FTP.microsoft.com (3) 0 Switched Ethernet Switched Ethernet www.microsoft.com (3) Switched Ethernet Switched Ethernet www.microsoft.com (3) search.microsoft.com (1) premium.microsoft.com (1) European Data Center FTP Download Server (1) SQL SERVERS (2) Router msid.msn.com (1) MOSWest Admin LAN SQLNet Feeder LAN FDDI Ring (MIS4) Router www.microsoft.com (5) Building 11 Live SQL Server Router home.microsoft.com (5) FDDI Ring (MIS2) www.microsoft.com (4) activex.microsoft.com (2) search.microsoft.com (3) register.microsoft.com (2) msid.msn.com (1) search.microsoft.com (3) register.microsoft.com (2) msid.msn.com (1) FDDI Ring (MIS3) www.microsoft.com (3) premium.microsoft.com (1) msid.msn.com (1) FDDI Ring (MIS1) www.microsoft.com (4) premium.microsoft.com (2) FDDI Ring (MIS3) www.microsoft.com (3) premium.microsoft.com (1) msid.msn.com (1) FDDI Ring (MIS1) www.microsoft.com (4) premium.microsoft.com (2) register.microsoft.com (2) msid.msn.com (1) Primary Gigaswitch Secondary Gigaswitch register.microsoft.com (2) msid.msn.com (1) Primary Gigaswitch Secondary Gigaswitch Staging Servers (7) search.microsoft.com (3) support.microsoft.com (2) register.msn.com (2) 1997 Microsoft.Com Farm MOSWest DMZ Staging Servers Live SQL Servers SQL Consolidators Live SQL Servers SQL Consolidators Japan Data Center www.microsoft.com (3) premium.microsoft.com (1) HTTP Download Servers (2) Router search.microsoft.com (2) SQL SERVERS (2) msid.msn.com (1) premium.microsoft.com (1) HTTP Download Servers (2) Router search.microsoft.com (2) SQL SERVERS (2) msid.msn.com (1) FTP Download Server (1) Router Router Router Router Router Router Router Router FTP Download Server (1) Router Router Router Router Router Router Router Router Internal WWW SQL Reporting Internal WWW SQL Reporting home.microsoft.com (4) home.microsoft.com (3) home.microsoft.com (2) register.microsoft.com (1) support.microsoft.com (1) home.microsoft.com (4) home.microsoft.com (3) home.microsoft.com (2) register.microsoft.com (1) support.microsoft.com (1) Internet 13 DS3 (45 Mb/Sec Each) 2 OC3 (45Mb/Sec Each) 2 Ethernet (100 Mb/Sec Each) cdm.microsoft.com (1) FTP Servers Download Replication FTP.microsoft.com (3) 0
  • 32. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 43 Scaleable Systems Scale UP and Scale OUT Everyone does both. Choice is • Size of a brick • Clones or partitions • Size of a pack
  • 33. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 44 Everyone scales out What’s the Brick? 1M$/slice • IBM S390? • Sun E 10,000?  100 K$/slice • Wintel 8X  10 K$/slice • Wintel 4x 1 K$/slice • Wintel 1x
  • 34. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 45 Clones: Availability+Scalability Some applications are • Read-mostly • Low consistency requirements • Modest storage requirement (less than 1TB) Examples: • HTTP web servers (IP sprayer/sieve + replication) • LDAP servers (replication via gossip) • App/compute servers or firewalls Replicate app at all nodes (clones) Spray requests across nodes. Grow by adding clones Fault tolerance: stop sending to dead clone. Growth: add a clone.
  • 35. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 46 Facilities Clones Need Automatic replication • Applications (and system software) • Data Automatic request routing • Spray or sieve Management: • Who is up? • Update management & propagation • Application monitoring. Clones are very easy to manage: • Rule of thumb: 100’s of clones per admin
  • 36. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 47 Partitions for Scalability Clones are not appropriate for some apps. • Statefull apps do not replicate well • high update rates do not replicate well • Huge DBs (disk to expensive to clone) Examples • Email / chat / … • Databases Partition state among servers Scalability (online): • Partition split/merge • Partitioning must be transparent to client.
  • 37. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 48 Partitioned (aka. Clustered) Apps Mail servers • Perfectly partitionable  Business Object Servers • Partition by set of objects. Parallel Databases Transparent access to partitioned tables Parallel Query
  • 38. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 49 Packs for Availability Each partition may fail (independent of others) Partitions migrate to new node via fail-over • Fail-over in seconds Pack: the nodes supporting a partition • VMS Cluster • Tandem Process Pair • SP2 HACMP • Sysplex™ • WinNT MSCS (wolfpack) Cluster In A Box now commodity Partitions grow in packs.
  • 39. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 50 What Parts+Packs Need Automatic partitioning (in dbms, mail, files,…) • Location transparent • Partition split/merge • Grow without limits (100x10TB) Simple failover model • Partition migration is transparent • MSCS-like model for services Application-centric request routing Management: • Who is up? • Automatic partition management (split/merge) • Application monitoring.
  • 40. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 51 Services on Clones & Partitions Application provides a set of services If cloned: • Services are on subset of clones If partitioned: • Services run at each partition System load balancing routes request to • Any clone • Correct partition. • Routes around failures.
  • 41. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 52 Farm pairs: Always Up  Two farms  Changes from one sent to other  When one farm fails other provides service  Masks • Hardware/Software faults • Operations tasks (reorganize, upgrade move • Environmental faults (power fail)
  • 42. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 53 Availabilty for a simple web site Web Clients Front End Web File Store SQL Temp State SQL Database Packs for availability Clones for availability Load Balance
  • 43. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 54 Cloned Packed file servers Packed Partitions: Database Transparency Farm Scale Out Scenarios SQL Temp State Web File StoreA Cloned Front Ends (firewall, sprayer, web server) SQL Partition 3 The FARM: Clones and Packs of Partitions Web Clients Web File StoreB replication SQL Database SQL Partition 2 SQL Partition1 Load Balance
  • 44. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 55 Clients IIS Web Server or other IP based services Data Servers SQL, Exchange, File Network Load Balancing Clones Cluster Service Pack COM+ Components Component Load Balancing (COM+) Clones Application Servers Reliable, Scalable, Modular 1 2 32 3 … 1 2 8 … 4 3
  • 45. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 56 Talk 2 (if there is time) Terminology for scaleability Farms of servers: • Clones: identical  Scaleability + availability • Partitions:  Scaleability • Packs  Partition availability via fail-over Farm Clone Partition Pack
  • 46. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 57 0 50 100 150 200 250 300 350 400 450 NT4 InProc W2K B1 inProc W2K RC1 InProc W2K RC2 InProc 1P 2P 4P 8P 0 50 100 150 200 250 300 350 400 450 NT4 InProc W2K B1 inProc W2K RC1 InProc W2K RC2 InProc NT4 OOP W2K B1 OOP W2K RC1 OOP W2K RC2 OOP 1P 2P 4P 8P Scalability: COM+ progress serving 1,000-statement ASP’s (servelets) Poor SMP Scaleability on IIS4 NT4 Big improvements from standard Transaction Processing tricks Out of Proc (safe execution) now much faster than In Proc was on IIS4 SPS: servelets per second (ASPs served per second by IIS, 1,000 statement VBscript) Shift from 4x200 Mhz to 8 450 Mhz
  • 47. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 58 Scaleability: So, What about the death of NT/Alpha? Two simultaneous Compaq TPC-C numbers  Intel Profusion  NT/SQL/COM+  550 Mhz  8 Processors  4 GB memory  40,368 TPM-C @ 18.46$/tpmC  745 K$ 5-year cost  Avail: 12/31/99  Alpha  Unix/Sybase/Tuexdo  700 Mhz  8 Processors  16 GB memory  42,437 TPM-C @ 55.45 $/tpmC  $2.35 M$ 5-year cost  Avail: 10/18/99 200% more expense for 5% more performance?
  • 48. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 59 Outline Summary of what you heard. (10 min) The database scene in general. (10 min) Scaleability: Farms, Clones,Parts & Packs (15 min) Microsoft DB research focus. (15 min) • TerraServer (design and ops). • RAGS. • Data Mining Q&A (10 min)
  • 49. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 60 The TerraServer http://www.terraserver.microsoft.com/
  • 50. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 61  Coverage: Range from 70ºN to 70ºS today: 35% U.S., 1% outside U.S.  Source Imagery: • 4 TB 1sq meter/pixel Aerial (USGS - 60,000 46Mb B&W- 151Mb Color IR files) • 1 TB 1.56 meter/pixel Satellite (Spin-2 - 2400 300 Mb B&W)  Display Imagery: 200x200 pixel images, subsample to build image pyramid  Store 5x compressed data  Nav Tools: • 1.5 m place names • “Click-on” Coverage map • Expedia & Virtual Globe • Pick of the week 1.6x 1.6 km “city view” .8 x .8 km 8m thumbnail .4 x.4 km browse 200x200 m tile Concept: User navigates an ‘almost seamless’ image of earth Database & application UI
  • 51. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 62 browser HTML Java Viewer The Internet Web Client Image Delivery Application SQL Server 7 Microsoft Site Server EE Internet Information Server 4.0 Image Provider Site(s) TerraServer DB Terra-Server Stored Procedures Internet Information Server 5.0 Image Server Active Server Pages (ADO) MTS TerraServer Web Site Software: Classic 3 Tier Design SQL Server 7 24 20 (8/12) 46 46 Fire wall
  • 52. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 63 SourceMeta Image Imagery Image Type Search Image Search Logical Schema Scale Job Load Job Load Mgmt External Link External Group External Geo Famous Category Famous Place TerraServer Terra Database TerraAdmin Admin Gazetteer Country Name State Name Place Name Place Type Small Place Name Pyramid
  • 53. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 64 TerraServer File Group Layout Convert 324 disks to 28 RAID5 sets plus 28 spare drives Make 4 NT volumes (RAID 50) 595 GB per volume Build 30 20GB files on each volume DB is File Group of 120 files E: F: G: H:
  • 54. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 65 Hardware 100 Mbps Ethernet Switch DS3 Site Servers Internet Map Server SPIN-2 Web Servers 2.9 TB Database Server AlphaServer 8400 8x400. 10 GB RAM 324 StorageWorks disks 10 drive tape library (STC Timber Wolf DLT7000 )
  • 55. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 66 Backup and Recovery • STK 9710 Tape robot • SQL Server Backup & Restore + • Legato Networker • Fast, incremental, differential, online • Clocked at 80 MBps (peak) (~ 200 GB/hr) Restore • Fast, incremental (file oriented), not online. Load & Backup&Recovery Performance Data Bytes Backed Up 1.2 TB Total Time 7.25 Hours Number of Tapes Consumed 27 Total Tape Drives 10 Data ThroughPut 168 GB/Ho Average ThroughPut Per Device 16.8 GB/Hour Average Throughput Per Device 4.97 MB/Sec NTFS Logical Volumes 2
  • 56. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 67 BAD OLD Load DLT Tape “tar” Drop’N’ DoJob Wait 4 Load LoadMgr DB 100mbit EtherSwitch 108 9.1 GB Drives Enterprise Storage Array Alpha Server 8400 108 9.1 GB Drives 108 9.1 GB Drives STC DLT Tape Library 60 4.3 GB Drives Alpha Server 4100 ESA Alpha Server 4100 LoadMgr DLT Tape NT Backup ImgCutter Drop’N’ Images 10: ImgCutter 20: Partition 30: ThumbImg 40: BrowseImg 45: JumpImg 50: TileImg 55: Meta Data 60: Tile Meta 70: Img Meta 80: Update Place ... LoadMgr
  • 57. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 68 New Image Load and Update ODBC Tx TerraLoader ODBC TX TerraServer SQL DBMS DLT Tape “tar” Metadata Load DB Active Server Pages Cut & Load Scheduling System Image Cutter Merge ODBC Tx Dither Image Pyramid From base
  • 58. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 70 TerraServer Daily Traffic Jun 22, 1998 thru June 22, 1999 0 10M 20M 30M Count Sessions Hit Page View DB Query Image After a Year: 1 TB of data 750 M records 2.3 billion Hits 2.0 billion DB Queries 1.7 billion Images sent 368 million Page Views 99.93% DB Availability 3rd design now Online Built and operated by team of 4 people TotalTime (Hours) 0 720 1440 2160 2880 3600 4320 5040 5760 6480 7200 7920 8640 Up Down Time (Hours:minutes) 0:00 0:30 1:00 1:30 2:00 2:30 3:00 3:30 4:00 4:30 5:00 5:30 6:00 Scheduled HW+Software Operations
  • 59. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 74 What TerraServer Shows Can serve huge databases on Internet for about a penny a page view mostly phone bill (!). Advertising pays more than a penny a page. Commodity tools do scale fairly far. A few people (3 developers, 1 operator) using power tools can build an impressive web site
  • 60. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 75 Outline Summary of what you heard. (20 min) The database scene in general. (10 min) Scaleability: Packs & Mobs (10 min) Microsoft DB research focus. (15 min) • TerraServer (design and ops). • RAGS. • Data Mining Q&A (15 min)
  • 61. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 76 Automatic Testing 60% of Microsoft R&D is testing. What can research do to help? • beyond joining the 500,000 Win2K beta testers Test generation robot: • Make up SQL queries • Send them to SQL Server, Oracle, DB2, Informix,… • If answer is the same, great, if not there is a problem Also good for stress tests Found MANY bugs in our products (all fixed). Found MANY bugs in other’s products. Very valuable tool.  MSR-TR-98-21 Massive Stochastic Testing of SQL, Slutz, Don http://research.microsoft.com/scripts/pubDB/pubsasp.asp?RecordID=175 W X Y Z 1672 1672 1672 1672 232 234 241 31 1 1 1 1 31 15 12 28 1 12 5 116 0 29 32 4 18 18 19 25 45 19 18 113 Error All four agree 84% Problem with intermediate table. Case W,X, and agree 95%
  • 62. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 77 Some Tera-Byte Databases Kilo Mega Giga Tera Peta Exa Zetta Yotta The Web: 1 TB of HTML TerraServer 1 TB of images Several other 1 TB (file) servers Hotmail: 20 TB of email Sloan Digital Sky Survey: 40 TB raw, 2 TB cooked EOS/DIS (picture of planet each week) • 15 PB by 2007 Federal Clearing house: images of checks • 15 PB by 2006 (7 year history) Nuclear Stockpile Stewardship Program • 10 Exabytes (???!!)
  • 63. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 78 Data Mining Find interesting structure (patterns, relationships) in data • Prediction • Segmentation (clustering) • Dependency modeling (find distribution) • Summarization • Trend and change detection and modeling Allow user to state the query in terms of the business logic • User does not speak statistics or SQL Use data to build predictors • regression, classification, segmentation etc. Generate summaries and reports for insight • find “easy to describe” segments in data automatically • find segments not known to analyst
  • 64. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 79 Data Mining: Microsoft SiteServer Commerce 3.0 Intelligent Cross-sell Based on: • Historical sales baskets in stores • Contents of current shopper basket • Browsing behavior of shopper Predict: ranking of products in store likely to be most interesting to shopper. Http://www.holtoutlet.com/outlet4
  • 65. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 80 Mail to 25% and capture 40% 400% improved response! 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 0.1% 0.2% 0.3% 0.6% 1.3% 5.3% 6.7% 25.5% 34.5% 43.8% 56.9% 68.5% 94.8% 98.5% 100.0% % mailed % Captured of true targets Real data drawn from a Microsoft marketing example
  • 66. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 81 How do people use www.microsoft.com? Data Mining (Clustering) Engine User browsing data 100M hits per day 14M users/week X segments Cluster Visualizer Wizard
  • 67. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 82
  • 68. Jim Gray, Research and Microsoft SQL Server Microsoft Research http://research.Microsoft.com/~gray/talks/ 3 October 1999 Chicago, Ill. 83 Outline Summary of what you heard. (10 min) The database scene in general. (10 min) Scaleability: Farms, Clones,Parts & Packs (15 min) Microsoft DB research focus. (15 min) • TerraServer (design and ops). • RAGS. • Data Mining Q&A (10 min)

Editor's Notes

  1. 3 October 1999
  2. 3 October 1999
  3. 3 October 1999
  4. 3 October 1999
  5. 3 October 1999
  6. 3 October 1999
  7. 3 October 1999
  8. 3 October 1999
  9. 3 October 1999
  10. 3 October 1999
  11. 3 October 1999
  12. 3 October 1999
  13. 3 October 1999
  14. 3 October 1999
  15. 3 October 1999
  16. 3 October 1999
  17. 3 October 1999
  18. 3 October 1999
  19. 3 October 1999
  20. 3 October 1999
  21. 3 October 1999
  22. 3 October 1999
  23. 3 October 1999
  24. 3 October 1999
  25. 3 October 1999
  26. 3 October 1999
  27. 3 October 1999
  28. 3 October 1999