3. Agenda In-Memory DBMS !! Altibase, TimesTen, ExtremeDB, KDB+, C-ISAM … Why ?! Ultra & Extreme low latency AND Exture+ How ?! NUMA, SSD, Infiniband, Compiler, MQ, Tick, CEP … Best ?! In-Memory Computing ?! Best Of Best !! High Performance Computing !!
4. In-Memory Architecture
5. In-Memory Database System High Performance Low Latency No Jitter Disk-Based RDBMS vs Oracle TimesTen ( Simple Architecture )
6. Modern DBMS ≒In-Memory DBMS
7. In-Memory Database System 4G 3G 2G 1G
8. Altibase HDB & XDB장점 국내개발 및 기술지원 증권사 레퍼런스 MVCC 지원단점 성능한계 & 메모리 이슈 고급 엔지니어 부재
9. Oracle TimesTen & Coherence장점 오라클 & 레퍼런스 성능 DA ( Direct Attach )단점 DA ( Direct Attach ) 한계 MVCC 미지원 Durability 이슈 엔지니어 및 기술지원 부제
10. McObject ExtremeDB & FE장점 성능 해외 통신 & 증권 레퍼런스 STAC Member Join단점 임베디드 전문 회사 국내 기술지원 FE ( Financial Edition ) 검증
11. KX Systems KDB+장점 WORLD BEST QL ( Q Language ) STAC-M3 Leader단점 가격 국내 기술지원
12. IBM Informix C-ISAM장점 OLD BEST 성능 & 가격 Zero Configuration & Admin단점 File DB ACID 미지원 유지보수 및 기술지원
13. second ㎳ ㎳~㎲ ㎲ 2 digit ㎲~㎱
14. Ultra & Extreme Low Lantecy
15. Exture+ RFP
16. Exture+ RFP
17. Exture+ RFP
18. Exture vs Exture+
19. Sync vs Async
20. Pub-Sub Architecture
21. Exture+ Dev ?!
22. Exture+ Dev ?!
23. Exture+ Dev ?!
24. Exture+ Dev ?!
25. NUMA Architecture
26. NUMA Architecture Using McObject’s 64-bit eXtremeDB-64, the application creates a 1.17 Terabyte, 15.54 billion row database on a 160-core Linux based SGI® Altix® 4700 server.
27. NUMA ArchitectureIt supports up to 512 * sockets or 1024 cores under one instance ofLinux and as much as 128TB of globally addressablememory.
28. NUMA Architecture CPU-Socket-Isolation via PhysicalNIC / PhysicalCPU pairing. Multiple CPU sockets holding multiple CPUs should be used like multiple machines. Avoid inter-CPU communication
30. InfiniBand & 10GbEIn computer networking, Server Message Block (SMB), also known as CommonInternet File System (CIFS, /ˈsɪfs/) operates as an application-layer network protocolmainly used for providing shared access to files, printers, serial ports, andmiscellaneous communications between nodes on a network.
31. InfiniBand & 10GbE
33. SSD is very very difficult.SSD (DRAM/NVRAM/flash/SLC/MLC) is Blah Blah !!
34. Compiler & *.[so|a] Library
35. Message QueueZeroMQ Crazy fast Brokerless architecture In-process library Lower latencies Very simple to use No persistence – requiring higher layers to manage persistenceRabbitMQ VMware vFabric AMQP compliant Written in erlang Small footprint and seemingly fewer lines of code in comparison to other AMQP compliant queue managers
36. OpenMAMA is high performanceMiddleware Agnostic Messaging API
37. Tick & Time series databaseeXtremeDB Financial Edition meets the specialized requirements ofhandling market data with several powerful features:Flexible data layout. eXtremeDB Financial Edition implements columnar datalayout for fields of type ‘sequence’. Sequences can be combined to form a time series, idealfor working with tick streams, historical quotes and other sequential data. The technologysupports database designs that combine row-based and column-based layouts, to bestleverage L1/L2 cache speed.Traditional DBMSs bring rows of data into L1/L2 cache for processing. But financial data –such as trades and quotes – are better handled by a column-based layout that avoidsflooding the cache with unwanted data.Vector-based statistical function library. Vector-based statisticalfunctions provide high efficiency by executing over all or part of one or more sequences andsupporting assembly lines of operations on sequences, for statistical/quantitative analysiseXtremeDB Financial Edition provides a rich library of vector-basedstatistical functions thatexecute over sequence to accelerate management of time series data. Handles real-time and historical data. eXtremeDB Financial Edition’sin-memory storage is ideal for real-time data, while developers can easily specify persistenttables for historical data with a simple notation in the database schema. eXtremeDBprogramming skills are fully interchangeable between in-memory and persistent databasedesigns (developers needn’t learn two database system products for real-time and historicaldata).