Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.
Copyright © 2015 Oracle and/or its affiliates. All rights reserved. |
DB12c on SPARC M7
In-Memory and Oracle SPARC M7
Patr...
Copyright © 2015, Oracle and/or its affiliates. All rights reserved. |
Why is Oracle doing
this
Copyright © 2015, Oracle and/or its affiliates. All rights reserved. |
In-Memory Accelerates Key Business Processes
Deep A...
Copyright © 2015, Oracle and/or its affiliates. All rights reserved. |
Real-Time Enterprise
100‘s
GB/s ETL (GB/s)
Data War...
Copyright © 2015, Oracle and/or its affiliates. All rights reserved. |
Lower Cardinality Data can be Most Interesting Data...
Copyright © 2015, Oracle and/or its affiliates. All rights reserved. |
Data Cardinality and Number of Bits to Encode
# of ...
Copyright © 2015, Oracle and/or its affiliates. All rights reserved. |
Lower Cardinality Data is Important to Most Analysi...
Copyright © 2015, Oracle and/or its affiliates. All rights reserved. |
How does it work
Copyright © 2015, Oracle and/or its affiliates. All rights reserved. |
Row Format Databases vs. Column Format Databases
Ro...
Copyright © 2015, Oracle and/or its affiliates. All rights reserved. |
SPARC
Oracle Database In-Memory Dual Format Databas...
Copyright © 2015, Oracle and/or its affiliates. All rights reserved. |
Oracle In-Memory Columnar Technology
• Pure in-memo...
Copyright © 2015, Oracle and/or its affiliates. All rights reserved. |
Why is an In-Memory scan faster than the buffer cac...
Copyright © 2015, Oracle and/or its affiliates. All rights reserved. |
Why is an In-Memory scan faster than the buffer cac...
Copyright © 2015, Oracle and/or its affiliates. All rights reserved. |
In-Memory Compression Unit (IMCU)
• Contiguous stor...
Copyright © 2015, Oracle and/or its affiliates. All rights reserved. |
Column Format Enables Hardware Acceleration
• Oracl...
Copyright © 2015, Oracle and/or its affiliates. All rights reserved. |
How does it impact
to environments
Copyright © 2015, Oracle and/or its affiliates. All rights reserved. |
Complex System is Slowed by Analytic Indexes
• Most...
Copyright © 2015, Oracle and/or its affiliates. All rights reserved. |
Column Store Replaces Analytic Indexes
• Fast analy...
Copyright © 2015, Oracle and/or its affiliates. All rights reserved. |
Test Case
(Load&Storage Data Scenario)
Have we test...
Copyright © 2015, Oracle and/or its affiliates. All rights reserved. |
Proof of Concept – what is the concept to proof
20
...
Copyright © 2015, Oracle and/or its affiliates. All rights reserved. | Oracle Confidential – Internal/Restricted/Highly Re...
Copyright © 2015, Oracle and/or its affiliates. All rights reserved. |
Proof of Concept – what is the concept to proof
PoC...
Copyright © 2015, Oracle and/or its affiliates. All rights reserved. |
Software in Silicon: Improving Performance
Reduce p...
Copyright © 2015, Oracle and/or its affiliates. All rights reserved. |
Decompress at memory speed >120 GB/sec
Software in ...
Copyright © 2015, Oracle and/or its affiliates. All rights reserved. |
Database In-Memory Acceleration Engines
• New SPARC...
Copyright © 2015, Oracle and/or its affiliates. All rights reserved. |
• SQL made up from few basic ops:
Filter/Search/Sor...
Copyright © 2015, Oracle and/or its affiliates. All rights reserved. |
Test Case
(Load&Storage Data Scenario)
What was in
...
Copyright © 2015, Oracle and/or its affiliates. All rights reserved. |
Methodology
28
• Throughput tests inspired on: ESG ...
Copyright © 2015, Oracle and/or its affiliates. All rights reserved. |
Methodology
29
• JMeter used to simulate users traf...
Copyright © 2015, Oracle and/or its affiliates. All rights reserved. |
Test Case
(Load&Storage Data Scenario)
Did we achie...
Copyright © 2015, Oracle and/or its affiliates. All rights reserved. |
SPARC M7:
Throughput and Response Time for 10,20,30...
Copyright © 2015, Oracle and/or its affiliates. All rights reserved. |
SPARC M7:
Response Time (ms) for 10,20,30,40,50,60,...
Copyright © 2015, Oracle and/or its affiliates. All rights reserved. |
Intel® Xeon® X5670 :
Throughput and Response Time f...
Copyright © 2015, Oracle and/or its affiliates. All rights reserved. |
Intel® Xeon® X5670 :
Response Time(ms) for 10,20,30...
Copyright © 2015, Oracle and/or its affiliates. All rights reserved. |
Intel® Xeon® E5-2699 v3 :
Throughput and Response T...
Copyright © 2015, Oracle and/or its affiliates. All rights reserved. |
Intel® Xeon® E5-2699 v3 :
Throughput and Response T...
Copyright © 2015, Oracle and/or its affiliates. All rights reserved. |
SPARC M7:
How to be sure that DAX is being used
Bec...
Copyright © 2015, Oracle and/or its affiliates. All rights reserved. |
How to make it even
faster ?
Copyright © 2015, Oracle and/or its affiliates. All rights reserved. |
Data Warehouse
Traditional OTLP & Analytics
• Key b...
Copyright © 2015, Oracle and/or its affiliates. All rights reserved. |
• Avoid evaluating predicates against every
column ...
Copyright © 2015, Oracle and/or its affiliates. All rights reserved. |
SPARC M7 Decompression+Scan & Range Scan
Dcompress ...
Copyright © 2015, Oracle and/or its affiliates. All rights reserved. |
SPARC M7 In-memory Advantages
SPARC M7 DAX is 1% of...
Copyright © 2015, Oracle and/or its affiliates. All rights reserved. |
Oracle In-Memory Columnar Technology
Scan via softw...
Copyright © 2015, Oracle and/or its affiliates. All rights reserved. |
Oracle In-Memory Columnar Technology
DAX scan proce...
Copyright © 2015, Oracle and/or its affiliates. All rights reserved. |
Summary
45
SPARC M7 shows significantly better resu...
Copyright © 2015, Oracle and/or its affiliates. All rights reserved. | Oracle Confidential – Internal/Restricted/Highly Re...
Oracle DB In-Memory technologie v kombinaci s procesorem M7
Upcoming SlideShare
Loading in …5
×

Oracle DB In-Memory technologie v kombinaci s procesorem M7

462 views

Published on

Webinář Oracle DB In-Memory technologie v kombinaci s procesorem M7
Prezentuje: Patrik Plachý, Oracle
23.3.2016

Published in: Technology
  • Be the first to comment

Oracle DB In-Memory technologie v kombinaci s procesorem M7

  1. 1. Copyright © 2015 Oracle and/or its affiliates. All rights reserved. | DB12c on SPARC M7 In-Memory and Oracle SPARC M7 Patrik Plachý Senior Consultant CoreTech Competency Center Oracle Confidential – Internal 1
  2. 2. Copyright © 2015, Oracle and/or its affiliates. All rights reserved. | Why is Oracle doing this
  3. 3. Copyright © 2015, Oracle and/or its affiliates. All rights reserved. | In-Memory Accelerates Key Business Processes Deep Analysis requires many passes through >10TBs of Data • In-memory transforms key business processes – Payroll processing (HCM) – Demand processing & demand sensing (SCM) – Material planning (ERP) – Financial close & virtual close (Financial) – Campaign management (CX) • Simultaneous Analytics & OLTP – Minimizes scheduled shutdowns for reports – Up-to-the-minute accurate analysis 3 Business Processes Transformed In-Memory 30x FASTER than flash storage! Dual-format: On-disk & In-memory consistent Column format SPARC In-memory on-disk Row format Typical fast IO only 40GB/s 1200‘s GB/s SPARC M7-8
  4. 4. Copyright © 2015, Oracle and/or its affiliates. All rights reserved. | Real-Time Enterprise 100‘s GB/s ETL (GB/s) Data Warehouse Server Real-Time Enterprise – Up-to-the-minute Analytics Analytics at memory speed – 1000’s GB/s is much greater old 1-40 GB/s bottlenecks • Key business processes radically transformed by orders magnitude – Costs also dramatically reduced on-disk Row format Millions IOPs on-disk Data Warehouse on-disk Typical fast IO 40GB/s Column format In-memory SPARC
  5. 5. Copyright © 2015, Oracle and/or its affiliates. All rights reserved. | Lower Cardinality Data can be Most Interesting Data More compression with lower cardinality(fewer bits needed), performance also increases • Analytics often distills data by grouping according to combinations of features – Again most interesting features have few dictionary Bits 5 Unique or Random Data Gender Season 5-point Scale Marital Status Top10 10 ranking Month Hour State Weeks Minutes Age Test Score Country US City >100k Days Top 500 Job Classification Area code Nasdaq NYSE Top 5,000 School districts zipcode DOB last 150 years Temperature RainfallWind direction Region Price Delivery status Most Interesting data has fewer dictionary bitsMake Model 2-bit 3-bit 4-bit 5-bit 6-bit 7-bit 8-bi t 9-bit 10-bit 11-bit 12-bit 13-bit 14-bit 15-bit 16-bit 17-bit 18-bit 19-bit... Cardinality n-bits (calculated as 2^n-bits)
  6. 6. Copyright © 2015, Oracle and/or its affiliates. All rights reserved. | Data Cardinality and Number of Bits to Encode # of unique column entries is cardinality, dictionary bits needed is log2(cardinality) • Sample of Common attributes: – Analytics distills interesting data into classes of features • Which means low Dictionary Bits are most interesting and most common Oracle Confidential – Highly Restricted 6 Attribute Cardinality Bits Attribute Cardinality Bits gender 2 1.0 Temparatue 200 7.6 Seasons 4 2.0 US cities >100,000 295 8.2 5-point scale 5 2.3 Wind direction 360 8.5 marital status 5 2.3 days 365 8.5 day of week 7 2.8 top 500 500 9.0 top 10 10 3.3 US Job Classifications 840 9.7 ranking 10 3.3 area code 999 10.0 months 12 3.6 Rainfall (tenths) 1,000 10.0 GPA levels 14 3.8 top 1,000 1,000 10.0 hours 24 4.6 #NYSE listings 1,867 10.9 US states 50 5.6 #NASDAQ listings 3,400 11.7 weeks 52 5.7 top 5,000 5,000 12.3 minutes spent 60 5.9 top 10,000 10,000 13.3 60 month 60 5.9 school districts 14,000 13.8 test score 100 6.6 top 50,000 50,000 15.6 age 150 7.2 DOB last 150 years 54,750 15.7 countries 195 7.6 zip codes 99,999 16.6
  7. 7. Copyright © 2015, Oracle and/or its affiliates. All rights reserved. | Lower Cardinality Data is Important to Most Analysis Online Retail Company • Lower cardinality tables common – Attributes of transactions often have fewer values (cardinality) – ex: part color, state, ratings, demographic data, etc. • Nearly unique data is 18 to 19-bits – Ex: index, customer ID, transaction ID, etc. SPARC M7 fast across spectrum Low Cardinality Medium Cardinality High Cardinality 1-4 5-8 9-12 13-16 17-19
  8. 8. Copyright © 2015, Oracle and/or its affiliates. All rights reserved. | How does it work
  9. 9. Copyright © 2015, Oracle and/or its affiliates. All rights reserved. | Row Format Databases vs. Column Format Databases Rows Stored Contiguously  Transactions run faster on row format – Example: Query or Insert a sales order – Fast processing few rows, many columns Columns Stored Contiguously  Analytics run faster on column format – Example : Report on sales totals by region – Fast accessing few columns, many rows SALES SALES 9 Until Now Must Choose One Format and Suffer Tradeoffs
  10. 10. Copyright © 2015, Oracle and/or its affiliates. All rights reserved. | SPARC Oracle Database In-Memory Dual Format Database • BOTH row and column formats for same table • Simultaneously active and transactionally consistent • Analytics & reporting use new in-memory Column format • New Analytics Compression means huge amounts of database can now fit in-memory • OLTP uses proven row format 10 Memory Memory SALES SALES Row Format Column Format
  11. 11. Copyright © 2015, Oracle and/or its affiliates. All rights reserved. | Oracle In-Memory Columnar Technology • Pure in-memory column format • Not persistent, and no logging • Quick to change data: fast OLTP • Enabled at table or partition • Only active data in-memory • 2x to 20x compression typical • Available on all hardware platforms 11 SALES Pure In-Memory Columnar
  12. 12. Copyright © 2015, Oracle and/or its affiliates. All rights reserved. | Why is an In-Memory scan faster than the buffer cache? SELECT COL4 FROM MYTABLE; 12 X X X X X RESULT Row Format Buffer Cache
  13. 13. Copyright © 2015, Oracle and/or its affiliates. All rights reserved. | Why is an In-Memory scan faster than the buffer cache? SELECT COL4 FROM MYTABLE; 13 RESULT Column Format IM Column Store RESULT X X X X X
  14. 14. Copyright © 2015, Oracle and/or its affiliates. All rights reserved. | In-Memory Compression Unit (IMCU) • Contiguous storage per column in an IMCU • CUs store Min/Max for some operations • Multiple formats: – For example, Dictionary Compression: CU stores (smaller) dictionary IDs instead of full values – Additional compression also possible • Dictionary encoding uses cardinality of Column to construct dictionary – 50 US are stored in only 6 bits (<1 byte) – Spelling out the state name is much longer • The name “South Dakota” needs 12 characters or 24 unicode bytes (192 bits vs 6 bits) 14 0 1 3 2 0 2 3 Column CU Min: South Dakota Max: Utah Dictionary Query Low (Dict) Column value list South Dakota Tennessee Utah Texas South Dakota Texas Utah Dictionary VALUE ID South Dakota 0 Tennessee 1 Texas 2 Utah 3
  15. 15. Copyright © 2015, Oracle and/or its affiliates. All rights reserved. | Column Format Enables Hardware Acceleration • Oracle 12c only scans needed columns • Processing values simultaneously – SPARC DAX Software in Silicon • SPARC M7 up to 220 Billion rows/sec • Scans offloaded from cores • M7 DAX scans data directly from memory – Intel AVX instructions: 256-bit registers • X86 E5 v3 up to 40 Billion rows/sec • Scans consumes cores (4-8 threads max BW) • Scanned data comes into caches • Memory bandwidth is critical factor – Single-thread performance does not predict in-memory performance SPARC M7 designed to accelerate in-memory scans & other in-memory DAXorx86SIMD Load Multiple Values At same time Simultaneously compare multiple values CPU Memory CA CA CA CA Example: Find all sales in state of CA STATE
  16. 16. Copyright © 2015, Oracle and/or its affiliates. All rights reserved. | How does it impact to environments
  17. 17. Copyright © 2015, Oracle and/or its affiliates. All rights reserved. | Complex System is Slowed by Analytic Indexes • Most Indexes in complex OLTP (e.g. ERP) databases are only used for analytic queries • Inserting one row into a table requires updating 10-20 analytic indexes: Slow! • Indexes only speed up predictable queries & reports Table 1 – 3 OLTP Indexes 10 – 20 Analytic Indexes 17
  18. 18. Copyright © 2015, Oracle and/or its affiliates. All rights reserved. | Column Store Replaces Analytic Indexes • Fast analytics on any columns • Better for unpredictable analytics • Less tuning & administration • Column Store not persistent so update cost is much lower • OLTP & batch run faster Table 1 – 3 OLTP Indexes In-Memory Column Store 18
  19. 19. Copyright © 2015, Oracle and/or its affiliates. All rights reserved. | Test Case (Load&Storage Data Scenario) Have we tested
  20. 20. Copyright © 2015, Oracle and/or its affiliates. All rights reserved. | Proof of Concept – what is the concept to proof 20 • Thanks to new SPARC dedicated acceleration engines built on chip scalability and performance of processing data with InMemory option should give much better results than compared to other CPU platform (in this case Intel®)
  21. 21. Copyright © 2015, Oracle and/or its affiliates. All rights reserved. | Oracle Confidential – Internal/Restricted/Highly Restricted 21 Servers
  22. 22. Copyright © 2015, Oracle and/or its affiliates. All rights reserved. | Proof of Concept – what is the concept to proof PoC DB In-Memory Acceleration Decompression Engines Application Data Integrity Sub-microsecond Cluster Messages Software in Silicon Performance Reliability Capacity Communication
  23. 23. Copyright © 2015, Oracle and/or its affiliates. All rights reserved. | Software in Silicon: Improving Performance Reduce processing time by off-loading simple tasks to special purpose hardware 23 Software processing Hardware processing Software processing Processing time Hardware processing Application performance improves because software processing supported by Software in Silicon is processed by hardware Without Software in Silicon With Software in Silicon
  24. 24. Copyright © 2015, Oracle and/or its affiliates. All rights reserved. | Decompress at memory speed >120 GB/sec Software in Silicon: Accelerating Oracle Database 12c One step faster Decompress More than Doubles data size Read Software scan Read Write Write Read DAX Write Multiple steps SQL: SELECT count(*) …WHERE lo_orderdate = d_datekey …AND lo_partkey = 1059538 AND d_year_monthnum BETWEEN 201311 AND 201312; t
  25. 25. Copyright © 2015, Oracle and/or its affiliates. All rights reserved. | Database In-Memory Acceleration Engines • New SPARC chip uses dedicated acceleration engines built on chip – Independently process streams of unaligned database column elements of any size • E.g. find all values that match ‘penguins’ • Frees CPU cores to run higher level SQL functions • Reads data directly from memory and places results in cache for core consumption – Shared cache provides ultra-fast communication Core Shared Cache SPARC CPU Core Core Core DB Accel DB Accel DB Accel DB Accel
  26. 26. Copyright © 2015, Oracle and/or its affiliates. All rights reserved. | • SQL made up from few basic ops: Filter/Search/Sort or Join/Group/Aggregate • First generation DAX (Query pipe) accelerates • Translate: HASH JOINs • Scan: search (“WHERE” clause) • Select: filter to reduce a column • Decompression more important than compression • Reading outweighs writing • Accelerate RLE, N gram, OZIP DAX DB Acceleration in High Performance Kernel Decompression and Query (Query Pipeline of DAX) DAX Engine(s) Core Core CPU
  27. 27. Copyright © 2015, Oracle and/or its affiliates. All rights reserved. | Test Case (Load&Storage Data Scenario) What was in a methodology?
  28. 28. Copyright © 2015, Oracle and/or its affiliates. All rights reserved. | Methodology 28 • Throughput tests inspired on: ESG Lab Review • Tests were issued with following criteria: • Identical schemas with the same size were generated on both platforms within a single database instance •Data generated with SSB (https://github.com/electrum) - SCALEFACTOR=100 • Instance Caging used to reference to 6 oracle db cpu licenses (resource_manager_plan=DEFAULT_PLAN): • M7 cpu_count = 96 -> 12 cores • Intel® Xeon® X5670 Processors (2.93 GHz): cpu_count = 24 -> 12 cores • Intel® Xeon® E5-2699 v3 Processors (2.3 GHz) cpu_count = 24 -> 12 cores • All data populated for in-memory • Tables compressed with MEMCOMPRESS FOR QUERY HIGH for in-memory • DOP used with DEGREE 8 on LINEORDER table TABLE_NAME INMEMORY INMEMORY_COMPRESS NUM_ROWS ------------------------------ -------- ----------------- ---------- CUSTOMER ENABLED FOR QUERY HIGH 3000000 DATE_DIM ENABLED FOR QUERY HIGH 2556 LINEORDER ENABLED FOR QUERY HIGH 600037902 PART ENABLED FOR QUERY HIGH 1400000 SUPPLIER ENABLED FOR QUERY HIGH 200000
  29. 29. Copyright © 2015, Oracle and/or its affiliates. All rights reserved. | Methodology 29 • JMeter used to simulate users traffic with following settings: • simulation iterations for simultaneous users 10,20,30,40,50,60,70,80,90,100 • following query was used: select count(distinct(lo_custkey)) from( select lo_custkey from lineorder,date_dim where lo_orderdate = d_datekey and d_weeknuminyear = :1 and d_year = :2 and lo_orderpriority <> :3 and lo_ordtotalprice between 7000 and 150000 group by lo_orderkey, lo_custkey having count(lo_linenumber) =1); • Bind variables values randomly picked from external .csv file • Performance of InMemory processing analyzed with two metrics: Query Throughput (tps) and Query Response Time (ms)
  30. 30. Copyright © 2015, Oracle and/or its affiliates. All rights reserved. | Test Case (Load&Storage Data Scenario) Did we achieve good results
  31. 31. Copyright © 2015, Oracle and/or its affiliates. All rights reserved. | SPARC M7: Throughput and Response Time for 10,20,30,40,50,60,70,80,90,100 users
  32. 32. Copyright © 2015, Oracle and/or its affiliates. All rights reserved. | SPARC M7: Response Time (ms) for 10,20,30,40,50,60,70,80,90,100 users
  33. 33. Copyright © 2015, Oracle and/or its affiliates. All rights reserved. | Intel® Xeon® X5670 : Throughput and Response Time for 10,20,30,40,50 users
  34. 34. Copyright © 2015, Oracle and/or its affiliates. All rights reserved. | Intel® Xeon® X5670 : Response Time(ms) for 10,20,30,40,50 users
  35. 35. Copyright © 2015, Oracle and/or its affiliates. All rights reserved. | Intel® Xeon® E5-2699 v3 : Throughput and Response Time for 10,20,30,40,50,60,70,80,90,100 users
  36. 36. Copyright © 2015, Oracle and/or its affiliates. All rights reserved. | Intel® Xeon® E5-2699 v3 : Throughput and Response Time for 10,20,30,40,50,60,70,80,90,100 users
  37. 37. Copyright © 2015, Oracle and/or its affiliates. All rights reserved. | SPARC M7: How to be sure that DAX is being used Because: • there are no DAX-specific wait events on database level to verify DAX usage • there are no DAX-specific statistics on database level to verify DAX usage Dtrace is the answer: • one can trace libdax_query.so.1 libdax.so.1 usage • use system level fbt provider (provides probes associated with the entry to and return from most functions in the Solaris kernel) it provides modules and functions for DAX Another way to trace DAX usage is busstat: -physical CPU counters that can be used to determine the performance of DAX pipelines. More details: http://blog.ora-600.pl/2016/01/27/oracle-sparc-m7-tracing-dax-with-dtrace-and-busstat/
  38. 38. Copyright © 2015, Oracle and/or its affiliates. All rights reserved. | How to make it even faster ?
  39. 39. Copyright © 2015, Oracle and/or its affiliates. All rights reserved. | Data Warehouse Traditional OTLP & Analytics • Key business processes slowed by traditional design – Often working on “old” business data ETL (GB/s) Data Warehouse Server Many operations bottlenecked on IO & delays in updates between separate serversMany operations bottlenecked on IO & delays in updates between separate servers on-disk Row format Millions IOPs on-diskon-disk Typical fast IO 40GB/s OLTP Server
  40. 40. Copyright © 2015, Oracle and/or its affiliates. All rights reserved. | • Avoid evaluating predicates against every column value – Check range predicate against min/max values • As before, skip IMCUs where min/max disqualifies predicate – If min/max indicates all rows will qualify, no need to evaluate predicates on column values Min $4000 Max $7000 Min $8000 Max $13000 Min $13000 Max $15000 Example: Find stores with sales between $8000 and $14000 NO ROWS Skip IMCU SOME ROWS Needs evaluation ALL ROWS Skip Evaluation Predicate Optimization: Reduce Predicate Evaluations ? 40
  41. 41. Copyright © 2015, Oracle and/or its affiliates. All rights reserved. | SPARC M7 Decompression+Scan & Range Scan Dcompress & Scan tasks occur in a single operation on SPARC M7 • SPARC M7 DAX offloaded acceleration frees cores & common ops in one step – Decompress & scan in one step • Both RLE & Ozip decompression – Range comparisons in one step • “How many between start-date and end-date?” 41 Decompress More than Doubles data size One step in SPARC hardware MemoryCompute Read MemoryCompute Read Software scan SQL: SELECT count(*) …WHERE lo_orderdate = d_datekey …AND lo_partkey = 1059538 AND d_year_monthnum BETWEEN 201311 AND 201312; time Multiple steps in Software DAX 10X Faster Read Write
  42. 42. Copyright © 2015, Oracle and/or its affiliates. All rights reserved. | SPARC M7 In-memory Advantages SPARC M7 DAX is 1% of chip that can do the work of tens of cores • DAX offloads the cores – DAX avoids cache pollution by not scanning with cores • DAX decompresses data at same rate as scan-only • DAX performs one-step range scans • SPARC M7 2x to 3x memory bandwidth for In-memory 42 SQL: select sum(lo_extendedprice*lo_discount) as revenue from lineorder, date_dim where lo_orderdate = d_datekey and d_year = 2012 and lo_quantity between 6 and 25 and lo_discount between 1 and 3 Processes: Decode values (DAX) & Sum aggregation (cores) Hash Joins (cores) Bloom Filter Joins (DAX & cores) Scans (DAX) Range Scans (DAX) Analytics M7 cores freed for OLTP DAX DAX DAX: Database Accelerator
  43. 43. Copyright © 2015, Oracle and/or its affiliates. All rights reserved. | Oracle In-Memory Columnar Technology Scan via software(including SIMD) consumes all of the cores 43 SALES (COMPRESSED)
  44. 44. Copyright © 2015, Oracle and/or its affiliates. All rights reserved. | Oracle In-Memory Columnar Technology DAX scan processing in DAX frees most of the cores for transactional processing 44 SALES (COMPRESSED)
  45. 45. Copyright © 2015, Oracle and/or its affiliates. All rights reserved. | Summary 45 SPARC M7 shows significantly better results in terms of Throughput and Response Time for InMemory operations. Thanks to Data Analytics Accelerators (DAX), which are in-memory query acceleration engines performance is many times faster when compared to other processors . It is noticeable for InMemory operations also in compressed format (DAX is able to operate directly upon compressed IMCUs). This makes SPARC M7 possible to perform real time analysis and fulfill business demands.
  46. 46. Copyright © 2015, Oracle and/or its affiliates. All rights reserved. | Oracle Confidential – Internal/Restricted/Highly Restricted 46

×