X

Alejandro Vargas' Blog

  • December 20, 2006

RAC and ASM Implementation Documents and Guides

Alejandro Vargas
Technical Leader, ACS Global Delivery, Infrastructure & BigData

For RAC 10g R2 using ASM implementation please follow these Guidelines:

RAC:

RAC site: http://www.oracle.com/technology/products/database/clustering/index.html
Step by Step Install Guide: 
Patch 10.2.0.3 include fix for Virtual IP not failing over if NIC is stopped.

Network Configuration:
Use 1GB interconnect over UDP with switches, crosscable is not supported.
Virtual IP Configuration Assistant (vipca), that is run automatically at the end of the RAC install, may fail with error : "the given iterface(s) eth0 is not public" if the IP range is within the range generally reserved for private networks, like 192.168.* and 10.10.*. See Note 316583.1

Storage:
ASM using ASMLib on Linux is the best practices.
OCFS2 for shared Oracle Homes, OCR, Voting Disk and shared ASM spfile
OCR, Voting Disk and shared ASM spfile can be located on Raw Devices also, in this case you need to setup boot scripts to set ownership, permissions, and start raw devices service.

ASM:
ASM Site: http://www.oracle.com/technology/products/database/asm/index.html
Best Practices: http://www.oracle.com/technology/products/database/asm/pdf/asm_10gr2_bptwp_sept05.pdf

Backup of ASM based databases:
Rman level 0 once a week + incremental level 1 daily for small / medium databases
If storage is EMC or Netapp, ASM is compatible with SRDF, Snapshots and other storage replication technologies
find relevant documentation on this site: http://www.oracle.com/technology/products/database/asm/index.html

EMC related papers:
http://www.oracle.com/technology/products/database/asm/pdf/asm-on-emc-5_3.pdf
http://www.oracle.com/technology/products/database/asm/pdf/EMC%20TF-SRDF%2002-06_0.pdf






Join the discussion

Comments ( 8 )
  • Kevinc Closson Wednesday, December 20, 2006
    Why eth0 for private?
  • Alejandro Vargas Thursday, December 21, 2006
    If you use eth0 for the virtual IP, vipca may abort with error "the given iterface(s) eth0 is not public"
    I've seen this happening in several installs where network people configured eth0 as public, known to DNS and eth1 as private, undefined on DNS.
    You still can run vipca manually and set virtual IP's on eth0 and will work without any problem.
    Is just to avoid the installer crashing at this point.
  • nm Sunday, December 24, 2006
    is this a known issue - i will be deploying rac and asm on hp-ux with serviceguard. I haven't made mention to the o/s guys if eth0 or lan0 should be public or private? By known i mean - is this a platform specific issue that has been known to occur everytime - or is it just one of those things?
  • nm Sunday, December 24, 2006
    i plan on doing a deployment on hp-ux, from what i recall (looking at the documentation, since i have not done this yet) hp-ux uses interface names like lan0 etc. So is this issue something specific to a platform or to just using eth0?
  • Alejandro Vargas Sunday, December 24, 2006
    What I've described was the behaviour I've seen in several RAC installs on Linux, RH3 and RH4 with 10g R2. I'll try to find if someone has seen this on other platforms.
  • Alejandro Vargas Sunday, December 24, 2006
    I did correct the information related to use eth0 or eth1, as I did find that the vipca problem is related to range of IP's and not to the network card interfaces. Please check Note 316583.1
  • nm Wednesday, March 7, 2007
    Hi, I know this post is a bit old. But was wondering if you've ever had to deploy 10gRAC R2 with asm as the storage option shared between two rac databases.? Typically there are 2 diskgroups one for data and the other for recovery files. Since We will have two databases I was wondering whether 4 diskgroups is a feasible way to go - 2 data groups 1 for each database and 2 diskgroups for recovery files - 1 for each database?
  • Alejandro Vargas Wednesday, March 7, 2007
    Yes I had a RAC environment with 2 databases sharing the same disk groups, we had datadg and archdg disk groups.
    This is fine if you are doing rman backups, that permit to backup each database separately; or consolidated backups of the two databases using srdf, bcv, or other storage snapshots; as you must backup both databases together with each split.
    Having separate disk groups for each database gives you more flexibility in terms of storage backups and I think is the best option.
Please enter your name.Please provide a valid email address.Please enter a comment.CAPTCHA challenge response provided was incorrect. Please try again.