Comparing the TCO of HP NonStop with Oracle RAC
1 like1,861 views
The document discusses the importance of high availability and fault tolerance in building mission-critical infrastructure, particularly focusing on HP Nonstop systems and Oracle RAC. It outlines scalability, data integrity, and recovery options, emphasizing the necessity for business applications to be resilient and perform effectively. Key concepts include various availability levels, the impact of failures, cost considerations, and configuration recommendations for data protection and system reliability.
Comparing the TCO of HP NonStop with Oracle RAC
- 1. © Copyright 2014 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. Better TCO& HigherAvailability thanOracleRAC with HP NonStop Azucena Ubierna HP NonStop PreSales GTUG 2015 azucena.ubierna@hp.com
- 2. © Copyright 2014 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. Business is extremely cost sensitive How should we build Mission Critical Infrastructure? Growing demand for Always ON Open Source SW is cheap .. even under EE distributions
- 3. © Copyright 2014 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. Maximize Data Protection Minimize Downtime - Mission Critical
- 4. © Copyright 2014 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. Maximize Data Protection OLTP Minimize Downtime - Mission Critical
- 5. © Copyright 2014 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. Impact of Failures - IDC Availability Spectrum HP NonStop systems are AL4 Impact of component failure AL4 AL3 AL2 AL1 Availability Level 3 Short outage is needed for failover to take place Availability Level 1 Need to switch to redundant resources before processing resumes. Availability Level 2 Balancing may not be perceptible to end-users because of retry NonStop delivers Source: IDC, September 2012, Doc #236946 Worldwide and U.S. High-Availability Server 2012-2016 Forecast and Analysis Availability Level 4 Switch to alternate resources is not perceptible to end users
- 6. © Copyright 2014 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. IDC Availability Level 4 SMP Enterpirse Servers Superdome Server built for resiliency. One big SMP box but can have isolated failure areas. Sysplex IDC Level 4 Server built for resiliency. Two big SMP boxes and SW is resiliency aware Exadata IF used in PAIRs IIDC Level 4 Server built for scalability. Several SMP boxes to run the DB SW & serveral SMP boxes to drive the access to storage Fault Tolerant Servers MPP HP NonStop IDC Level 4 Server built for Fault Tolerance. Full SW stack is HW & SW Fault Tolerance aware Stratus, NEC.. IDC Level 4 Server builf for Fault Tolerance OS SW is HW Fault Tolerance aware. Guest SW is not Virtual Servers Most vendors provide High Availability & Fault Tolerant options. Same names but not necessarily same meanings Standard Server Clusters Stand by Clustering Servers built for standard resiliency. Two SMP boxes, one active and one standby. Farm Clustering Servers built for standard resiliency. All active boxes. When failure happens activity needs to be redirected. Mission Critical Infrastructure – Availability Spectrum
- 7. © Copyright 2014 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. Access Granted to All Committed Business Operations DB Coherence & Logs ACID DBMS Atomicity Consistency Isolation Durability Extensive Sets of Tools for Dumps, Replica and Recovery BASE DBMS Basically Available Soft-state Eventual Consistency Typically left to whatever the underlying FS provides. Cold snapshots. Storage System ESS Standalone ESS Or with Sync Replicated ESS Compulsory for some clustered HA options: - Stand By - Multi node RDBMS RAID Directly Attached to Standalone Fault Tolerant Servers Directly Attached to Standard servers in Clustered Scale-Out “RAID” deployments Back Up Solution Simple Tape Drives sized for the required duty along with Catalog Management SW Fully integrated Enterprise Backup Systems that may include encryption, de-duplication, replica… DR Site - RTO/RPO requirements Cold RTO >> 0 & RPO >> 0 Minutes >> 10 Based on - Stand By resources - DB Restore - Restart of Services Warm RTO > 0 & RPO > 0 Minutes < 10 Most Widely used DB Replication Tools Hot RTO = 0 & RPO = 0 Latency - Manage Commits Dead Locks - Manage Access RTO = 0 & RPO >0 Consistency - Manage Ping Pong - Manage Collisions Ensuring Data Protection & Data Recovery Capability
- 8. © Copyright 2014 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. Scalability SMP Scale UP Mainframe Applications Mainframe “Standard” Applications Superdome Scale OUT Stateless Applications Cluster aware Applications MPP External Interconnect Cluster aware ACID DBMS Oracle RAC Cluster aware BASE DBMS Internal Interconnect Big Data Oracle App Teradata Netezza Exadata MPP aware applications HP NonStop Processing Power Scalability & DATA Scalability
- 9. © Copyright 2014 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. Scalability SMP Scale UP Mainframe Applications Parallel Sysplex Fault Tolerance Managed by OS “Standard” Applications Superdome Stand By Clustering HA Managed by Clustering SW Scale OUT Stateless Applications Cluster aware Applications MPP External Interconnect Cluster aware ACID DBMS Oracle RAC Cluster aware BASE DBMS Internal Interconnect Big Data Apps Oracle RAC Teradata Netezza Exadata * 2 Oracle RAC Managed HA NonStop Applications & DB HP NonStop Fault Tolerance Managed by OS Scalability & Availability
- 10. © Copyright 2014 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. Minimize the Impact of Failures Build Fault Tolerance for each and every layer Usage of redundant resource Smart provisioning Smaller areas of failure “All Active” Clustering “Stand-by” Clustering Eliminate Single Points of Failure Availability, Data Integrity & Scalability Desgin Principles and a few “Not-So-Easy-Tasks” Data Scalability ACID properties preservation DataBase BackUp OnLine Replica Multi-node DataBase Pros & Cons for OLTP Data Preservation and Recovery
- 11. © Copyright 2014 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. Mission Critical Infrastructure & Features Summary Scalability High Availabilty Data Integrity Recovery SMP HW Fork Lift & Node Addition Multiplicity Platform Dependent Fail Over or DR NonStop MPP HW Up Lift & Incremental Built-in At every component SF Take Over or DR Linux OS - Scale UP Passive – NUMA constrains Stand By Clustering Passive Fail Over or DR Linux OS – Scale OUT Passive – App/DB dependent SL Retry or Clustering Passive Take Over or DR NonStop OS Active Active End-To-End SF Take Over or DR Oracle S1, SE, EE One Node ( 2, 4, N sockets) Stand By Clustering Redo logs RMAN / Data Guard Oracle EE RAC Multi Node ( 3 nodes max ?) RAC Multi - Redo logs RMAN / Data Guard NonStop SQL/MX 16 Blade units per Node Built-In TMF – Audit Trail TMF / RDF Application HW Aware, State Less, ?? HW Aware, State Less, ?? DB Dependent DB Dependent
- 12. © Copyright 2014 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. Some Extra Payback Expected Provisioning IT Resources for Mission Critical OLTP Business Some Relevant Costs Added • HW Infrastructure & Basic SW stack: • High Availability • Capability to Scale & Perform • Data Base Services: • Data Integrity • Capability to Scale & Perform • Data Protection & Recovery • Business Apps: • Resilient & Scalable SW • Operations: • MC Staff, • Change Management … • HW & SW Infrastructure • Processing Capacity • Storage Resources • Basic SW Stack • Data Base Services
- 14. © Copyright 2014 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. MPP Architecture NonStop OS NonStop SQL/MX Load Balancer RAC DataBase Nodes: from 2 to..? Interconnect & Coherence External Storage Subsystem External Storage Subsystem Application Nodes: from 2 to N Stateless or Cluster Aware Integrated Storage Subsystem Integrated Storage Subsystem DB App DB App DB DB DB Scale-out SMP servers & MPP NonStop servers Two approaches to Availability, Scalability & Data integrity
- 16. © Copyright 2014 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. Pure StandBy Clustering at Each Site ISV suggested configuration includes DB Replica to DR Site Active DB Node @ Primary Site Replicates to Active DB Node @ DR Site Primary Site DR Site
- 17. © Copyright 2014 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. • OS, App, DB • RHEL Lic E-LTU • (No Compilers included) • ServiceGuard for Linux Advanced Edition • Oracle EE, Adv Sec & RAC • Operations & Management • HP Insight Control E-LTU • HP 3PAR 7200 Reporting Suite E-LTU . • Processing Power DB & App – 2 HP BL460 Gen8 1P 4 core 32GB RAM – 2 HP BL460 1P 4 core 32GB RAM • Storage Subsystem – 6 HP 82B PCIe 8Gb FC Dual Port HBA – 2 HP Virtual Connect FlexFabric, 10Gbps/24p – 1 HP 3PAR StoreServ 7200 2-N Storage Base – 24 HP M6710 300GB 6G SAS 10K HDD • Communication Subsystem – 2 HP Virtual Connect FlexFabric, 10Gbps/24p • Rack Elements – HP BLc7000 CTO 3 IN LCD Plat Enclosure – HP 642 1075mm Shock Intelligent Rack – HP R5KVA UPS 3U IEC309-32A HV Intl Kit Specific ISVs 100 to 200 tps rate - Generic HP c7000 Proliant BL460c 1 Proc q Core & 3PAR 7200
- 18. © Copyright 2014 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. • OS, App, DB • NONSTOP OS, MISSION CRITICAL EDITION • NS OS SECURITY ENHANCEMENT J SER • NONSTOP TS/MP (ACS) • C/C++ COMPILER, TNS/E NATIVE • NONSTOP SERVER FOR JAVA 6.0 • NONSTOP JDBC TYPE 4 DRIVER-100+USERS • NONSTOP SQL/MX ENGINE & TABLE • DR Synchronization • REMOTE DATABASE FACILITY/IMPX • NONSTOP AUTOSYNC SOFTWARE • EXPAND • Operations, Management & Performance • WEB VIEWPOINT • MEASURE • I/O ESSENTIALS - WINDOWS CONSOLE & HOST • NONSTOP CLUSTER ESSENTIALS-WINDOWS & HOST • NS CLUSTER ESSENTIALS - BLADES HOST & • NS PERF ESSENTIALS-HP SIM CLIENT & HOST • NS SOFTWARE ESSENTIALS R2 -WIN CONS & HOST • Processing Power – 2 HP NONSTOP CPUs (Pulson based) • 1 Core CAPABLE • 32GB RAM per CPU • Storage Subsystem – 26 146 GB 15K SAS 6G SFF HDD – 2 HP NONSTOP SAS HBA 6G R3 – 2 HP NONSTOP STORAGE CLIM R4 – 2 SAS 6G SFF ENCLOSURE • Communication Subsystem – VIO Pair with VIO extension • 2x6 Ehternet ports • Rack Elements – 1 HP NONSTOP SINGLE PHASE UPS INTL R2 – 1 NONSTOP EXTENDED RUN TIME MODULE R2 – Rack Mounted Console Extrapolation for a 100 tps rate from an ISV/Hp Benchmark HP Integrity NonStop NS2302
- 19. © Copyright 2014 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. Building Resiliency for DB nodes failures From Pure StandBy Clustering to Cluster AWARENESS
- 20. © Copyright 2014 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. NonStop to Linux 3y TCO comparison
- 22. © Copyright 2014 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. Building Minimal HA Features on top of ISV suggestion From NO HA to Pure StandBy Clustering
- 23. © Copyright 2014 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. • OS, App, DB • RHEL Lic E-LTU • (No Compilers included) • Oracle Min, Oracle EE and Oracle RAC • Operations & Management • HP Insight Control E-LTU • HP 3PAR 7200 Reporting Suite E-LTU • Processing Power DB & App – 2 HP DL385p Gen8 8-SFF CTO Server, 128 GB RAM – 1 HP DL385p Gen8 8-SFF CTO Server, 64 GB RAM(*) • Storage Subsystem – 6 HP 82B PCIe 8Gb FC Dual Port HBA – 2 HP 8/24 Base 16-ports Enabled SAN Switch – 1 HP 3PAR StoreServ 7200 2-N Storage Base – 8 HP M6710 300GB 6G SAS 15K 2.5in HDD – 8 HP M6710 100GB 6G SAS 2.5in SLC SSD • Communication Subsystem – 3 HP Ethernet 10GbE 530FLR-SFP+ FIO Adptr – 3HP Ethernet 10Gb 2P 530SFP+ Adptr • Rack Elements – HP 642 1075mm Shock Intelligent Rack – HP R5KVA UPS 3U IEC309-32A HV Intl Kit Specific ISVs 300 tps rate for Specific Customer HP Proliant DL385p clusters & 3PAR 7200
- 24. © Copyright 2014 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. • OS, App, DB • NONSTOP OS, MISSION CRITICAL EDITION • NS OS SECURITY ENHANCEMENT J SER • C/C++ COMPILER, TNS/E NATIVE • NONSTOP SQL/MX ENGINE & TABLE • Operations, Management & Performance • MEASURE • I/O ESSENTIALS - WINDOWS CONSOLE • I/O ESSENTIALS - BLADESYSTEM HOST • NONSTOP CLUSTER ESSENTIALS-WINDOWS • NS CLUSTER ESSENTIALS - BLADES HOST • NS PERF ESSENTIALS-HP SIM CLIENT • NS PERFORMANCE ESSENTIALS-NSK HOST • NS SOFTWARE ESSENTIALS R2 -WIN CONS • NS SOFTWARE ESSENTIALS R2 NB HST • Processing Power – 4 HP NONSTOP CPUs (Tukwila based) • 4 Core CAPABLE BLADE • 48GB RAM per BLADE • Storage Subsystem – 44 146 GB 15K SAS 6G SFF HDD – 2 HP NONSTOP SAS HBA 6G R3 – 2 HP NONSTOP STORAGE CLIM R4 – 2 SAS 6G SFF ENCLOSURE • Communication Subsystem – 2 HP NONSTOP IP CLIM 5C R4 – 2 HP NONSTOP IP CLIM 3C/2F R4 • Rack Elements – 2 HP NONSTOP SINGLE PHASE UPS INTL R2 – 1 NONSTOP EXTENDED RUN TIME MODULE R2 – Rack Mounted Console Extrapolation for a 300 tps rate from Specific Customer Production Environment HP Integrity NonStop Blade System NB54004
- 25. © Copyright 2014 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. Sizing for the Different tps rates on both Options (*) Just one App server in the picture and quotes
- 26. © Copyright 2014 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. NonStop to StandBy Linux - 3 year TCO comparison -
- 27. © Copyright 2014 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. Building Resiliency for DB nodes failures From Pure StandBy Clustering to Cluster AWARENESS
- 28. © Copyright 2014 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. NonStop to Linux RAC DB - 3 year TCO comparison -
- 30. © Copyright 2014 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. NonStop X to Linux RAC DB - 3 year TCO comparison -