1 / 43

Extreme Performance with the Oracle Exadata X3 and DB 12c

Extreme Performance with the Oracle Exadata X3 and DB 12c. 娄恒 Heng.Lou@Oracle.com Exadata Solution Specialist.

vail
Download Presentation

Extreme Performance with the Oracle Exadata X3 and DB 12c

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Extreme Performance with the Oracle Exadata X3 and DB 12c 娄恒 Heng.Lou@Oracle.comExadata Solution Specialist

  2. The following is intended to outline our general product direction. It is intended for information purposes only, and may not be incorporated into any contract. It is not a commitment to deliver any material, code, or functionality, and should not be relied upon in making purchasing decisions. The development, release, and timing of any features or functionality described for Oracle’s products remains at the sole discretion of Oracle.

  3. 传统数据库部署问题 Servers • 传统数据库架构无法很好的扩展 • 服务器与存储的分离、SAN存储和网络瓶颈 • 多核CPU对于数据处理能力10倍于SAN存储的数据吞吐能力 • 投资成本 • 服务器、存储、SAN交换机、集成、存储软件、数据库软件 • 人员成本和技术复杂度 • 部署复杂,需要每个系统调优、测试、维护 SAN Bottleneck Storage

  4. Exadata平台的设计理念 • 为最佳运行数据库平台而设计 • 对服务器、存储、网络的集成和优化 • 大规模服务器集群实现水平扩展 • Flash优化、存储节省 • 通过对存储、Flash、网络特定的工程化软件设计,带来数据库极限性能的提升 • 部署简单、维护简单 • 适用于各种数据库负载 • OLTP、数据仓库、数据库云

  5. 超过1000+大客户部署 一半数据仓库、一半OLTP和数据库云 • PB级数据仓库 • 互联网在线交易平台 • 电子商务网站 • 数百个数据库整合

  6. Exadata for ERP, CRM, HR E-business Suite, Siebel, Peoplesoft, JDE, SAP

  7. Exadata架构:水平扩展平台(Scale-Out) 运行数据库的完整平台– 服务器、存储、网络 • 水平扩展的数据库服务器 • 8台–每台2路服务器(16核)256 GB内存 • 或者2台– 每台8路服务器(80核)2 TB 内存 • 水平扩展的智能存储服务器 • 14台– 2路存储服务器(12核) • 168 磁盘– 500TB • 56 Flash PCI – 22 TB Flash • 高速网络 • 提供 40 Gb/sec的内部连接 • 10Gb 或 1Gb数据中心标准网卡连接 Fully Redundant

  8. Exadata X3 | 内存数据库 X3 2012 2009 2010 2008 大容量闪存 所有I/O都在闪存中 竖向扩展 在线交易和 超大数据库 数据仓库 磁盘数据库 内存数据库

  9. 从小配置起步,不断演进和扩展 可以跨代系扩展 满机柜 二分之一机柜 八分之一机柜 四分之一机柜

  10. 无缝的升级和扩展 升级例子 X3-2 2012从1/2升到满配 • 一台Exadata可以同时使用多个版本的服务器 • 数据库和集群可以跨越不同版本的硬件 • 新的软件可以运行旧的硬件上 X2-2 2011从1/4升级到1/2配 V2 2010最早买了个1/4配

  11. 不需其他设备情况下,可扩展到18个机柜 • 5900计算Core • 9 PB非压缩数据存储

  12. 高度工程化和标准化 更低风险、更好的效果 • 避免工程师花大量的时间在集成和优化系统上 • Database, OS, drivers, firmware all optimized together • 把I/T的精力转移到业务分析上 • 开箱即用、快速部署完成 • Oracle提供从磁盘到数据库的端到端支持 • 可以运行目前所有的数据库负载任务 Ready- to-Run

  13. 对整个Oracle组件提供更好服务 包括对数据库软件的更好服务 对错误的主动式的远程监控 业界领先的服务响应时间: 5 分钟错误通知 15 分钟恢复或者升级到开发团队 30 分钟和开发团队联合调试 Oracle 工程师提供季度提供的季度更新补丁服务 对Exadata特定配置 有效 Exadata白金服务 ORACLE 白金服务 提供高可靠保障. 没有附加成本.

  14. Exadata关键创新低成本下的极致性能 • 横向扩展的智能存储 • 通过InfiniBand 和服务器连接 • Smart Scan 技术把查询负载下移 • 混合列式压缩 • 10x 数据仓库压缩率 • 15x 存档数据压缩率 未压缩 + + + 当扫描和存在flash时,数据保持压缩状态 压缩 • 智能PCI 闪存 • 磁盘的前端的透明缓存 • 30倍提高随机I/O 性能 • 4倍提高数据扫描速度 主数据库 在后续使用中保持空间节省 开发 备用数据库 备份 测试

  15. Exadata智能存储网格 • 数据密集操作在Exadata存储网格中运行 • 从磁盘中过滤行和列 (168 Intel Cores) • 举例: How much product X sold last quarter • Exadata Storage Reads 10TB from disk • Exadata Storage Filters rows by Product & Date • Sends 100GB of matching data to DB Servers • 大量存储服务器并行执行,消除IO瓶颈 100GB/sec Data Throughput

  16. Exadata智能存储 • Exadata存储服务器运行大量复杂的计算操作 • Join filtering • Incremental backup filtering • I/O prioritization • Storage Indexing • Database level security • Offloaded scans on encrypted data • Data Mining Model Scoring • 一般情况可以减少10倍以上的数据量,返回给DB服务器计算 Exadata Intelligent Storage Grid smart scan efficiency =522:1 =cell physical IO bytes eligible for predicate offload/cell physical IO interconnect bytes returned by smart scan

  17. Exadata混合列压缩 Highest Capacity, Lowest Cost • 数据按照列来组织和压缩 • Speed Optimized Query Mode for Data Warehousing • 10X compression typical • Runs faster because of Exadata offload! • Space Optimized Archival Mode for infrequently accessed data • 15X to 50X compression typical Query Faster and Simpler Backup, DR, Caching, Reorg, Clone Benefits Multiply

  18. Real-World数据压缩比 • Compression Ratio varies by customer and table • Compression of largest table at 10 ultra large companies • Average revenue > $60 BB • Average Query Compression ratio was 13x • On top of Oracle’s already highly efficient format Retail Financial Telecom

  19. Turkcell: DW and DB Consolidation Pre-ExadataData Warehouse Exadata V2 Data Warehouse 2 Exadata X2-2Prod Original V2 Prod/DR/Dev Hitachi USP-V5 Racks EMC DMX-45 Racks Backup 2011 2010 Restore Faster Reports Storage Savings Data Center Cost Savings ReducedAdmin 2011 900 TB 10X 80% Less Power30 m2 Less Space 20% • 4 Prod, 2 Test Databases • 400 TB Raw/ 40 TB Compressed • 2 RAC Clusters • 250 TBRaw Data • 25 TB Compressed “In a word, Oracle Exadata is fantastic. Almost no report takes more than 10 minutes to run, versus hours before. It sounds unreal, but it’s real.” - Power User, Finance Department, Turkcell • 2 Prod Databases • 600 TB Raw / 60 TB Compressed • 16-node RAC Cluster 27 min to 3 min (avg for 50k rpts) 1,000 TB to 100 TB

  20. SK Telecom: Revenue Assurance DW Exadata V1Data Warehouse (2009) Exadata V2Data Warehouse Exadata V2+X2-2 Data Warehouse Exadata X2-2Qtr Rack (2010) 2010 2011 • Billing Analysis System • Billing Analysis System • Usage & Roaming Inquiries • 1,000 TB Raw / 100 TB Compressed • 16-node RAC Cluster • Spam Filtering Extreme Capacity Storage Savings Capture Lost Revenue 10 TB/day $ Millions 900 TB “Oracle Exadata Database Machine has overwhelmingly superior performance, with average data processing rates 8 to 20 times greater than other SK Telecom in-house systems.” - Jin-hyung Lee, Manager, Network Engineering Department, SK Telecom 1,000 TB to 100 TB

  21. Exadata X3 Database In-Memory Machine 2 or 4 TB DRAM • X3 分级内存机制提供极致性能 • 自动把所有活动数据从磁盘移动到内存 • DRAM 内存扩展到2 TB 或者4 TB , 存放最热数据 • 压缩后可存储4 TB 到40 TB 用户数据 • 闪存扩展4倍,达到每个机柜22 TB • 压缩后可存储40 TB to 200 TB 用户数据– 所有活动数据 • 在线交易系统: 每秒1.5 Million 的SQL 随机读 • 相当于15,000 磁盘,150磁盘柜 • 数据仓库和报表:每秒100 GB SQL 数据扫描 • 相当于1,000磁盘,10磁盘柜 最热数据 22 TB PCI FLASH 活动数据 500 TB DISK 冷数据

  22. 加速交易响应时间 Uses Flash for Database Logs in a clever way Flash is fast but has slow outliers - Erase cycles, wear leveling, etc. Smart Flash Log transparently uses Flash as a parallel write cache to disk controller cache Whichever write completes first wins (disk controller or flash) Faster log writes and more throughput, uses almost no flash space Exadata Smart Flash Log Default (on left)- Choppy Response- High Outliers Transaction Response Times Smart Flash Log Enabled Smart Flash Log- 3x fasterresponse- Much lower outliers Automatic and Transparent

  23. Exadata X3 vs. 传统存储阵列 Exadata . . . • Smart Storage • True Scale-Out • InfiniBand • PCI Flash • Achieved from SQL • Not Counting Compression… 100GB/sec 52GB/sec With Flash Disks

  24. Exadata 数据库云 HR • Exadata 具有同时独特的能力,可以在一个云环境中同时运行多个不同负载类型的数据库 • 高端OLTP, 数据仓库,跑批,报表,备份等 • 同一时间运行 • X3 database in-memory 为所有负载提供极致性能 • 同时避免一种负载把其他应用的性能拖垮 EDW CRM ERP Supply Chain 所有负载,所有应用 SAP, Siebel, PeopleSoft, JDE,E-business Suite, Fusion Applications Data Mart

  25. U.S. Customs and Border Protection Faster Applications Storage Savings CostSavings Maximum Availability 1 Petabyte 99.95% 75% Cost Reduction 10X speedup “Oracle Exadata has changed our strategic IT focus from building systems to developing and supporting application services critical to the fulfillment of our national security mission.” - Ken Ritchhart, Deputy Assistant Commissioner 32 billionqueries/day Replaced $$ SAN 2010 - 2011 2010 2011 - 2012 Exadata V2Primary/Standby Exadata X2-2Primary /Standby Exadata X2-8Primary/Standby Data Guard Data Guard Data Guard Automated Targeting Systems Automated Export Systems Passenger Systems Enterprise Data Warehouse

  26. Exadata I/O Resource Management 适用于混合负载和多数据库云环境 • Ensure different databases are allocated correct resources • Database A: 33% I/O resources • Database B: 67% I/O resources • Ensure different users and tasks within a database are allocated the correct resources • Database B: • Interactive: 60% of I/O resources • Batch: 40% of I/O resources HR EDW CRM ERP Supply Chain Data Mart

  27. Performance, availability, usage by databases, services, clusters Software and hardware alerts Topology view of DB systems/cluster Drill down from database issue to hardware root cause Enterprise Manager 12c Exadata软硬件端到端的管理平台

  28. Exadata最佳的数据库运行平台 • Best Platform for… • Data Warehousing • OLTP • Database Cloud OLTP with Analytics and Parallelism of Warehousing Warehousing with Interactivity, Availability, and Security of OLTP

  29. Oracle Database 12.1对Exadata OLTP优化 • Database 12.1 enhancements for Exadata are focused on OLTP and Consolidation • Exadata enhanced for 12.1 Multitenant databases • Provides extreme performance and safest consolidation • Online Compression Tiering and Storage Tiering • Hybrid Columnar Compression OLTP Enhancements • Smart scans on LOBs • Enhanced Availability and Migration • SQL Performance Enhancements

  30. Database 12.1 多租户数据库架构 理想的Database as a Service Single Multitenant Database Many Databases each in a OS & VM SalesDB AssetsDB BillingDB PartsDB AssetsPDB Sales PDB Billing PDB PartsPDB • 数据库架构转变,适应于数据库云整合 • 资源共享、应用数据库独立和隔离 • 动态供给和升级 • 完全应用透明,不需要应用修改 • 每个数据库更少的硬件资源占用 Single Physical Container Database O/S O/S O/S O/S Single O/S, No VMs Needed VM VM VM VM

  31. Exadata网络资源管理 Network Resource Management maintains steady performance • OLTP workloads can degrade when network intensive reports or jobs are run • This is an especially serious issue on an unpredictable consolidated system • Exadata Network Resource Management uniquely prioritizes critical database messages through the entire fabric • Latency sensitive messages prioritized over batch, reporting, and backup messages • Log file writes have highest priority to ensure low latency transactions

  32. OLTP自动压缩分层 Frequent Reporting Compliance & Reporting OLTP • Automatic Compression Tiering converts cooler partitions to columnar compressed format online • Global indexes maintained • HCC now has row level locking for better OLTP • Get OLTP performance of row format, with compression and reporting benefits of columnar format 10x compressed 15x compressed This Year’s Data Prior Years Last Year Row Format is best for OLTP Column Format is best for reporting & archiving

  33. 存储分层和迁移 • New command moves database files and tablespaces to a new location online • Enables moving to a different ASM disk group • Use cases: • Move tables or partitions to different storage tier • E.g. Historical tablespaces can move from high performance to high capacity disks • Move to different mirroring • E.g move from normal redundancy to high • Move to newer hardware

  34. LOB处理加强 • Smart Scan 支持 LOB数据 • Smart Scan offload for small CLOBs (< 4KB) • Like expressions, e.g. like(html, ‘Exadata’) • Pattern Matching, e.g. regexp_like() • Smart Scan works for compressed and encrypted CLOBs • Parallel DML 可以支持在non-partitioned LOB columns • Parallel Insert as Select, Create Table as Select, Move Table or Partition • Also now parallelizes within partitions

  35. 可靠性的加强 • 远程同步(Far Sync) • 支持远距离的0数据损失的Standby数据库 • 全球数据服务(Global Data Services) • 跨数据中心进行无缝的负载均衡和故障切换failover • 应用连续性(Application Continuity) • 屏蔽应用级错误,当数据库实例失败后,通过自动化应用在其他实例重做 Primary Active Standby Active Standby GoldenGate

  36. Exadata ASM可靠性加强 • 优先恢复关键文件 • Critical files like log files and control files are rebalanced first to reduce the chance of data loss due to double failures (e.g double disk failure) • Prioritizing important files is a big advantage over generic storage mirroring • 对RebalanceI/O操作卸载到Exadata存储服务器 • Data sent directly from cell to cell rather than through database node • 数据同步支持速度控制和断点续传 • Also periodically checkpoints so that failed resyncs restart at last checkpoint

  37. 减少数据迁移的复杂性和停机时间 原数据库 (Big Endian) 目标数据库 – Little Endian • Cross Platform Backup and Restore optimizes migration to Exadata • Endian conversion performed without staging migration files • Minimize offline or read-only impact with multiple incremental backups Restore Backup Backup

  38. Road Map: OLTP消息处理的硬件优化 • Exadata已经整体极大提升了存储和网络的吞吐能力 • Exadata将更进一步加速OLTP数据库服务器间的消息处理机制 • 重点提升在RAC Cache Fusion的消息传递的延时和吞吐能力 • 数据库直接调用InfiniBand硬件,绕开网络软件的堆栈、中端和调度等 3X Speedup

  39. 总结 • 第四代数据库云服务器 • X3 Database in-memory Machine • 所有活动数据存在内存中 • Exadata 数据库云 • 新的1/8配入门级Exadata X3

  40. 选择Exadata平台的各种理由 快速部署 更好的用户体验 低能耗 运行速度 降低实施风险 线性扩展 维护简单 更快、更强的执行力 降低成本和复杂性

More Related