Technical white paper Run Oracle OLTP workloads in HP-UX vPars and Integrity VM v6.1.5 HP-UX 11i v3 Table of contents Introduction 2 Acronyms and terms used in this paper 3 Use HP-UX vPar and Integrity VM for hardware consolidation 3 Oracle OLTP test environment (used during this performance characterization) Native and vPar/Integrity VM host configuration Storage configuration HP-UX vPar and Integrity VM v6.1.
Introduction Run your mission critical applications like Oracle in an HP Converged Infrastructure to provide a flexible platform for deployment of your virtualized enterprise applications for today and the next decade. HP Integrity systems combine years of trusted availability and resiliency with efficiencies of HP blade systems for the foundation of the world’s first mission-critical converged infrastructure.
Acronyms and terms used in this paper Acronyms/Terms Definition AVIO Accelerated Virtual I/O AWR Oracle Automated Workload Repository DB Database DCOE HP-UX Data Center Operating Environment DIO Direct I/O DSS Decision support system GUID Manager Globally Unique Identifier Manager FC Fibre Channel LVM HP Logical Volume Manager LUN Logical Unit Number NPIV N-Port ID Virtualization OLTP Online transaction processing PGA Oracle Program Global Area RAID Redundant array of inexpensiv
vPars supports dedicated CPU and memory resources, while Integrity Virtual Machines supports shared CPU and memory. In both cases, IO resources can be dedicated 1 or shared. Virtualization makes it easier to optimize hardware resources since one Integrity server can host multiple OS instances.
Oracle OLTP test environment (used during this performance characterization) Figure 1 illustrates the system configuration used during this benchmarking exercise. The configuration is described as completely as possible to enable others to reproduce the test environment and measurements. Figure 1.
Native and vPar/Integrity VM host configuration The Integrity BL890c i2 Server Blade is configured as follows: Hardware configuration • CPU: 8 Intel Itanium Processor 9350s (1.73 GHz, 24 MB) 32 cores (4 per socket) HT = OFF at OS level and ON at firmware level • Memory: 256 GB • Storage: One dual-port 8 Gb/s FC mezzanine card, each port connected to an 8 Gb/s FC interconnect module in an HP BladeSystem c7000 Enclosure. • Networking: One 10GbE LAN-on-motherboard (LOM) interface.
Storage configuration The 4400 EVA is configured as follows: • Two disk shelves, both fully populated with 146 GB 15k RPM disks • Dual active controllers, each with 2 GB of battery-backed cache Two disk groups are created to store Oracle data files and redo logs: • Oracle data disk group – 12 virtual disks evenly distributed between controller A and B • Oracle redo log disk group – 8 virtual disks evenly distributed between controller A and B HP-UX vPar and Integrity VM v6.1.
• All storage is presented to the guest using whole LUNs from EVA – 12 LUNs used for holding Oracle data files – 8 LUNs used for holding Oracle redo logs • Dual 8 GB Fibre Channel NPIV capable HBA Software configuration • HP-UX 11i v3 September 2012 Data Center OE • HP Integrity VM v6.1.5 guest kit • No additional patches were installed, as HP-UX 11i v3 September 2012 Data Center OE has all performance-related fixes • HP OnlineJFS 5.0.
• Automatic Memory Mode enabled • Archive Mode disabled • Three redo log groups, each 4 GB in size • Data files created with bigfile option to allow growth beyond 32 GB • init.ora Parameters – db_block_size=8192 (8 KB Database Block Size) – memory_target=38912M (38 GB Memory Target, it is advisable to have memory target size around or less than 60 percent of the guest memory) Workload generator configuration Swingbench is a workload generator for Oracle databases.
Figure 2. Swingbench Transaction Throughput screen snapshot Vary the workload characteristics Most Oracle databases experience different workload characteristics throughout the day, week, and month based on the number of users attached to the database instance batch processing demands, and so on. Two of the more common performance variables that change based on workload are the amount of physical disk I/O’s and the CPU utilization of the database server.
Generation of database An empty database instance is created using the Database Configuration Assistant (DBCA). DBCA creates the default data files, the redo log groups, and other miscellaneous files. The Datagenerator utility (part of the Swingbench suite) is used to populate the dataset based on a specified scaling factor. As indicated earlier, for this benchmark effort with 64 GB of guest memory, we selected a 38 GB database memory target.
Note: It is not recommended to run any application at the VSP level, therefore the “native” mode tests are performed on an HP-UX server that did not have the VSP configured. As mentioned earlier in this paper, the Integrity BL890c i2 Server Blade has more physical memory than was intended for the virtualized environment (256 GB vs. 64 GB); so, it was necessary to limit the blade’s visible memory during the native mode test runs, to provide a balanced comparison.
HP-UX vPars and Integrity VM best practices while running Oracle (observed during this performance characterization) Here are some best practices, observed when running an Oracle workload inside a vPars and an Integrity VM guest. HP-UX vPars and Integrity VM host best practices • Do not run any workload at the same level as the VSP host. The VSP host should only be used to run Integrity vPars or VM guests. • Tune the host specifically to run vPars and Integrity VM guests.
Oracle support support.oracle.com Swingbench dominicgiles.com Customer feedback HP welcomes your input. Please send us your comments about this white paper, or suggestions for related documentation, through our technical documentation feedback website: hp.com/bizsupport/feedback/ww/webfeedback.html. Get connected hp.com/go/getconnected Current HP driver, support, and security alerts delivered directly to your desktop © Copyright 2012 Hewlett-Packard Development Company, L.P.