User manual HP STORAGEWORKS ENTERPRISE FILE SERVICES CLUSTERED GATEWAY

[. . . ] HP StorageWorks Using Oracle® on Enterprise File Services Clustered Gateway user guide Part number: 416226­002 First edition: September 2007 Legal and notice information © Copyright 2007 Hewlett-Packard Development Company, L. P. Hewlett-Packard Company makes no warranty of any kind with regard to this material, including, but not limited to, the implied warranties of merchantability and fitness for a particular purpose. Hewlett-Packard shall not be liable for errors contained herein or for incidental or consequential damages in connection with the furnishing, performance, or use of this material. This document contains proprietary information, which is protected by copyright. [. . . ] If the Database Administrator wishes to install Oracle executables (e. g. , ORACLE_HOME, APPL_TOP) into the EFS Clustered Gateway, the Storage Administrator will need to mount the filesystem with regular mount options. As is the case with host-based filesystems (e. g. , Ext3, UFS, JFS, VxFS, etc. ), a Standard file optimized mount is suitable for non-RAC databases. DBA's may wish to use exported Standard file optimized mounts for non-RAC databases for performance reasons. Standard optimized mounted filesystems benefit from cache in the EFS Clustered Gateway NAS heads, whereas DB optimized mounted filesystems are not cached at all in the EFS Clustered Gateway NAS heads. Figure 1 shows the EFS Clustered Gateway Management Console Graphical User Interface where the filesystem called /u01 is being mounted on the EFS Clustered Gateway NAS heads called tmr6s7 and tmr6s8. Since this filesystem is intended for a shared Oracle Home installation with the standard Optimal Flexible Architecture (OFA), the Std. File Optimized button has been clicked. Figure 1 EFS Clustered Gateway Management Console filesystem mount dialog 14 Configuring Oracle on the EFS Clustered Gateway DB optimized mount option For Real Application Clusters databases and Clusterware files (e. g. , 10gR2 Clusterware OCR/CSS, 1 1g Clusterware OCR/CSS), the DB Optimized mount option is a requirement for performance reasons. Note that failing to mount filesystems with the DB Optimized mount option will cause performance degredation only for RAC instances. The NAS heads in the EFS Clustered Gateway are fully cache-coherent so there is no threat of data corruption should this administrative step be overlooked. Figure 2 shows another EFS Clustered Gateway Management Console screenshot where a filesystem called /u02 is being mounted with the DB Optimized mount option. This filesystem mount option is intended for storage of Oracle Real Application Clusters databases and Clusterware files. Figure 2 EFS Clustered Gateway Management Console for DB optimized filesystem mount operation EFS Clustered Gateway Virtual NFS Service export options The EFS Clustered Gateway requires the same NFS export options as most other OSCP-certified NAS offerings. The three required export options are: · no_wdelay · rw · no_root_squash Figure 3 shows an EFS Clustered Gateway Management Console screenshot with two filesystems (/u01, /u02) configured in a single Export Group called oracle, all with the required export options. For more information on Export Groups, please refer to the "Configure Export Groups" section in the HP StorageWorks Clustered File System administration guide or the "mx exportgroup - Export Group commands" section in the HP StorageWorks Clustered File System Command Line reference guide. HP StorageWorks Using Oracle on HP EFS Clustered Gateway user guide 15 Figure 3 EFS Clustered Gateway Management Console NFS export group options Hosting Virtual NFS Services The EFS Clustered Gateway provides a truly revolutionary NFS server implementation with special support for highly available NFS. All NFS export groups are presented to NFS clients via a Virtual NFS Service, or VNFS for short. Virtual NFS technology is very important for two reasons: · Failover--If a NAS head in the EFS Clustered Gateway fails, the Virtual NFS Services running on the failed nodes will be transparently failed over to a backup NAS head. The NFS clients and the processes with open file handles on the filesystems involved will not be affected in any way. · Re-hosting--Using the EFS Clustered Gateway Management GUI or CLI, the administrator can move an active NFS service from one NAS head to the other for such reasons as load balancing or maintenance. This operation is also fully transparent at the NFS client level and there is no need to stop applications. A VNFS is a combination of Virtual Host IP and a proprietary enhancement to the NFS Server stack to support completely transparent NFS client failover. The filesystems remain accessible without re-mounting and, most importantly, processes with active file handles accessing files in the NFS exports are not impacted. For in-depth information regarding Virtual NFS Services in the EFS Clustered Gateway, please refer to "Configure Virtual NFS Services" in the HP StorageWorks Clustered File System administration guide or to "mx vnfs - Virtual NFS Service commands" in the HP StorageWorks Clustered File System Command Line reference guide. Figure 4 shows the EFS Clustered Gateway Management Console GUI with an example of associating an Export Group with a Virtual NFS Service as well as the NAS heads that will act as the primary and 1st backup in the event of a NAS head failure. [. . . ] For example, on Linux, the following kernel parameters need to be set. The installation guide for 10gR2 x86_64 recommends the following settings: kernel. shmall = 2097152 kernel. shmmax = 2147483648 kernel. shmmni = 4096 kernel. sem = 250 32000 100 128 fs. file-max = 65536 net. ipv4. ip_local_port_range = 1024 65000 net. core. rmem_default = 262144 net. core. rmem_max = 262144 net. core. wmem_default = 262144 net. core. wmem_max = 262144 Based on testing, the following kernel tunables and their associated values are recommended when using Oracle on Linux NFS clients accessing the EFS Clustered Gateway: net. ipv4. ip_forward = 0 fs. file-max=327679 fs. aio-max-nr=3145728 vm. lower_zone_protection=100 net. ipv4. ipfrag_high_thresh=524288 net. ipv4. ipfrag_low_thresh=393216 net. ipv4. tcp_rmem=4096 524288 16777216 HP StorageWorks Using Oracle on HP EFS Clustered Gateway user guide 21 net. ipv4. tcp_wmem=4096 524288 16777216 net. ipv4. tcp_timestamps=0 net. ipv4. tcp_sack=0 net. ipv4. tcp_window_scaling=1 net. ipv4. conf. default. rp_filter = 1 net. ipv4. conf. default. accept_source_route = 0 net. ipv4. ip_local_port_range=1024 65000 net. core. rmem_default=524288 net. core. wmem_default=524288 net. core. rmem_max = 16777216 net. core. wmem_max = 16777216 net. core. optmem_max=524288 net. core. netdev_max_backlog=2500 sunrpc. tcp_slot_table_entries=128 sunrpc. udp_slot_table_entries=128 See the Oracle Installation Guide for your specific Oracle port for any operating system configuration issues for Oracle on NAS. 22 Configuring NFS clients for Oracle on the EFS Clustered Gateway 5 Configuring Oracle for the EFS Clustered Gateway It is important to configure Oracle on the NFS clients to use the O_DIRECT flag when it opens files in the Enterprise File Services Clustered Gateway. For Oracle ports that support Oracle on Network Attached Storage, the init. ora parameter filesystemio_options governs whether certain filesystem features will be used by the server. With NFS, most ports of Oracle Database 10g Release 2 support both direct and asynchronous I/O with NFS mounts. [. . . ]