Driving Behavioral Change for Information Management through Data-Driven Gree...
Clouds in Your Coffee Session with Cleversafe & Avere
1. Clouds in Your Coffee
Webcast Series
Welcome to the
We will begin in a
moment.
Connect to audio by phone or VOIP
(small download required.)
2. Presenters
2Proprietary & Confidential
Mark Eastman
Sr. Director of Business Development
Avere Systems
Russ Kennedy
Sr. VP of Product Strategy & Customer Solutions
Cleversafe
3. Ten Years of Software Innovation
supported by more than 150 patent awards
Hundreds of Petabytes installed
in large enterprise and massive scale customers
10101010
10101010
10101010
Venture-backed, strongly-financed
private company in Chicago
3
CleversafeCleversafe®®About
Build a great, enduring storage company.
mission:
4. The clean-sheet
solution for object
storage
The clean-sheet
solution for object
storage
leverage commodity
hardware, dramatically
lowering cost
Software-defined
to deliver enterprise-class
function and security at
Exabyte+ scale
Designed for scale
Founding
PrinciplesPrinciples
4
5. 5
CapabilitiesCapabilitiesUnique
Bit-perfect data integrity using
distributed erasure coding
24 X 7 uptime
No RAID, no replicationReliability
Architecture scales to Exabytes
Zettabytes target within 3 Years
No single point of failureScalability
Data at rest and in motion encryption
without key management
AD, LDAP, PKI authenticationSecurity
Performance scales linearly
SmartRead and SmartWrite capability
Performance optimizations for geo-
dispersed systemsPerformance
…At a fraction of the cost of Traditional Storage!
6. Efficiency
AdvantageAdvantage
How to build a highly reliable storage system
for 1 Petabyte of usable data?
RAID 6, Replication, Tape Cleversafe®
dsNet®
Original
1.20 PB Raw
Remote copy
1.20 PB Raw
Onsite mirror
1.20 PB Raw
567 TB Raw 567 TB Raw567 TB Raw
1 PB
3.6 PB
900
3.6x
3.6x
20.8 KW
3 FTE
Replication/backup
Usable Storage 1 PB
Raw Storage 1.7 PB
4TB Disks 432
Disks Required 1.7x
Floor Space 1.7x
Power Required 6.46 KW
Ops Staffing .5 FTE
Extra Software None
Tape backup
1 PB
$
65% +
TCO Savings
7. How it Works
7
Original Data
Accesser®
Appliance
Encrypted,
Erasure Coded,
Slices
Slicestor®
Appliances
Site 1 Site 2 Site 3
11 55 99
22 66 1010
33 77 1111
44 88 1212
Data is encrypted and erasure
coded, transforming the original
object into a number of data
“slices” such that 100% of the
object data can be read using a
subset of the created slices.
(Total slice created = width = N)
1
Each slice of a given object
is stored on a different
storage node. As a result, the
loss
of access to the data on one
storage node impacts at most
1 slice of a given objects data.
2
When requested, object reads are
fulfilled using a subset of the available
slices. The ability to provide 100%
of the object data from a subset of
its stored slices is the basis of data
reliability and availability. (Slices
required to read = threshold = K)
3
11 22 44 121299 1010 1111
Cleversafe ®
IDA
8. UseCasesUseCases
Cloud storage Active archive
Content repositories
and collaboration
Business
Continuity
Back-up repository
Unstructured content
distribution
Storage as a service
8
10. Avere Edge-Core Architecture
10Proprietary & Confidential
Edge Filer
• Performance optimized
• Dynamic tiering of active data
• Accelerate read, write & metadata operations
• Low latency access from fast media (RAM,
NVRAM, SSD, SAS)
• Linear performance scaling through clustering
• Responsible for data placement on Core filers
(write, move, mirror)
Core Filer
• Capacity optimized
• FlashCloudTM
support for public and private
object
- Public object: Amazon S3
- Private object: Cleversafe
• Any 3rd
-party NAS (e.g. NetApp, EMC/Isilon,
Oracle, BlueArc, Nexenta)
• Support for object and NAS Core filers from
heterogeneous vendors in the same cluster
Private ObjectPrivate Object
Core Filers
Public ObjectPublic ObjectWAN
WAN
Legacy NAS (local & remote)
Avere FXT Edge FilerClient
Workstations
Compute
Farm
Low latency read,
write & metadata
ops
Add performance
& capacity
Place data
11. Global Namespace
11Proprietary & Confidential
Global Namespace
• Single mount point for all Core filers
• Single namespace *and* accelerated performance
• Easy to add/remove Avere to/from environment
• Create junctions (e.g. /sw) for improved namespace mgmt
FlashMove®
• Non-disruptively move data (e.g. /audi) between Core filers
FlashMirror®
• Mirror write data (e.g. /src) to two locations for DR
/
/sw
/src
/hw
/camaro /audi
Global Namespace
Logical path
unchanged
/archive
/mktg
/assets
Private ObjectPrivate Object
Core Filers
Public ObjectPublic ObjectWAN
WAN
Legacy NAS (local & remote)
Avere FXT Edge FilerClient
Workstations
Compute
Farm
FlashMove
X
/camaro
/audi
/src
/assets
/archive
/src’
/audi
FlashMirror
/ on Avere
Single mount point
Write
12. Comparing 1,000,000 IOPS Solutions*
EMC Isilon
$10.7 / IOPS
NetApp
$5.1 / IOPS
150ms
Avere
$2.3 / IOPS
Throughput
( I OPS)
Latency/ORT
( ms)
List Price $/I OPS Disk
Quantity
Rack
Units
Cabinets Product Config
Avere FXT 3800 1,592,334 1.24 $3,637,500 $2.3 549 76 1.8
32- node cluster,
cloud storage config
NetApp FAS 6240 1,512,784 1.53 $7,666,000 $5.1 1728 436 12 24-node cluster
EMC Isilon S200 1,112,705 2.54 $11,903,540 $10.7 3360 288 7 140-node cluster
*Comparing top SPEC SFS results for a single NFS file system/namespace. See www.spec.org/sfs2008 for more information.
Avere 32-node
FXT cluster
Core Filer
-NAS
-Public object
-Private object
The next three slides describe the technical inner workings of the Avere Cloud NAS solution, showing “how we do it.”
Avere invented an architecture we call “Edge-Core” that is purpose-built for the cloud. In this architecture, the NAS filer is decomposed into two parts. First, there is an Edge filer that is optimized for performance and sits near the users of the storage. The Edge filer is provided by Avere. Second, there are one or more Core filers that are optimized for capacity and reside at a cloud service provider or on the enterprise premise. The Core filer can be public object storage (provided by a cloud storage service provide like Amazon), private object storage (provided by a vendor like Cleversafe), or legacy NAS (provided by a 3rd-party NAS vendor like NetApp, EMC/Isilon, Oracle, BlueArc, or Nexenta).
This slide includes some simple animation to emphasize a couple key points. (click)
The Edge flier provides low latency data access to the users. This is accomplished by three distinct features of the Edge filer: 1) Dynamic tiering automatically moves the active data from the Core filer(s) to the Edge filer. 2) The Edge filer is a full-fledged filer and locally terminates all read, write, and metadata operations. 3) All these operations are served out of fast media (RAM, SSD, and/or SAS) that compose the Edge filer.
Over time as the user and application demand at the Edge increases, the performance and capacity of the Edge filer can be scaled linearly by adding nodes to the FXT cluster.
In addition to delivering low latency and scalable performance out at the edge, the Edge filer is responsible for placing the data on the appropriate Core filer for long term data retention. Write data is automatically sent to the Core filers based on a schedule defined on the Edge filer. The Edge filer also has the ability to transparently move live data between Core filers and mirror data to multiple Core filers for redundancy.
When selecting the right Core filer for your application, your decision is all about what provides the best capacity (cost and density) and long-term data protection for your data. With the 4.0 release, Avere is introducing new FlashCloud software licenses that allow using public and private object storage as Core filers. Public object storage supported in 4.0 includes Amazon S3 and Glacier. Private object storage supported in 4.0 includes Cleversafe. Just as we have from the beginning, 4.0 also supports 3rd-party NAS solutions from the vendors shown on this slide and more. Public object, private object, and legacy NAS storage are added to the FXT cluster in a manner that is simple and transparent to the user. Under the covers, the FXT cluster speaks object APIs to the object storage and NFS to the NAS filers. Release 4.0 allows mixing storage types (public object, private object, and NAS) and storage vendors (Amazon, Cleversafe, NetApp, EMC) on the same FXT cluster.
Avere offers GNS functionality that simplifies data management and access in storage environment that have multiple storage systems in use (which is most of them). Avere is the only vendor to provide GNS that spans public object, private object, and legacy NAS, all from heterogeneous vendors.
With the Avere GNS, clients have a single mount point on the FXT cluster and from there can access all the data across all the systems. GNS solutions have gotten a bad rap from solutions like the F5/Acopia ARX product that used to be on the market because they had a negative impact on performance and made spaghetti of your file systems so they were hard to add and remove. Avere GNS has neither of this faults. On the contrary, Avere provides GNS *and* performance acceleration in the same product. And, the Avere FXT cluster is simple to add to and remove from existing environments.
With the GNS in place, we can provide transparent data migration with our FlashMove software. (CLICK) This enables moving data between Core filers for load balancing or adding new storage or removing old storage or moving data to an archive, etc. FlashMove physically moves the data between storage systems but the data remains in the same logical location due to our GNS so the movement is non-disruptive to clients and applications. In this animation, we are moving the hardware engineering sub-tree /audi to public object storage for archiving. As a side note, at Avere we name our hardware platforms after cars that our CEO has owned. Our first generation platform was called “camaro” and this went EOA a couple years ago so this slide shows /camaro already archived on public object storage. Our second generation hardware was called audi which more recently went EOA so that’s the one I demonstrated arching to public object storage in the timeframe of today.
We also provide mirroring with our FlashMirror software, which enables implementing DR. (CLICK) In this example our software source code in the sub-tree /src is stored on a local NAS system in the enterprise data center and FlashMirror creates a copy of this sub-tree in the public cloud. FlashMirror creates a baseline copy of /src from the primary to the secondary and then from that point forward all changes to /src are made on both the primary and secondary.
Avere has delivered performance acceleration and scaling to many different customers in many different industries and applications. This includes applications such as rendering, transcoding, seismic processing, genomic sequencing, financial simulations, software build, and more. This is a place in the presentation where you may want to insert a customer case study from our library that is relevant to the audience of your presentation. In the standard presentation we used the SPEC SFS benchmark since it is the most relevant workload across the broad range of customers we sell to.
In the world of file systems there is a well know benchmark called SPEC SFS that is used to compare the performance of NAS systems. All the NAS vendors use the benchmark and post their results on the website shown at the bottom of this slide. SPEC does a great job of providing a detailed, apples-to-apples comparison of NAS products running in a “typical” enterprise-class NAS environment.
This slides compares three top performance results for a single NFS file system/namespace on the SPEC site. Avere is the current record holder with almost 1.6 million ops/sec achieved on a 32-node FXT 3800 cluster. Note that this is not a max cluster from Avere. We used just enough nodes to achieve the top spot. Today we can go to 50 nodes per cluster and will go beyond this in the future. In second place is NetApp with a max 24-node cluster mode system. In third place is EMC/Isilon with a max 140-node system S-Series cluster.
While achieving the highest performance was an important point for Avere, our primary points were: 1) we can scale performance at the edge while consolidating the storage in the core over a WAN, and 2) our solution is the most efficient.
To make point 1), we inserted 150ms of latency between the FXT cluster and the Core filer. 150ms of latency is equivalent to the latency of a transcontinental WAN link (e.g. USA to Europe or USA to Japan).
To make point 2), just look at the sizes of the systems. We are faster than NetApp and Isilon in just a fraction of the space. 1.8 racks and 4 feet wide for us. 14 feet wide for Isilon. 24 feet wide for NetApp.
If you scan across the orange row you can see the details of our performance advantage and our higher efficiency. Avere is the highest performance, the lowest latency, the lowest cost, and use the least amount of space and power.