Diaster Tolerance with Serviceguard for Linux and Oracle 11g RAC

[Nguyen Van Hai]

Hi Everyone,

I am designing a Disaster Tolerance solution for my customer (FSI segment) by using HP EVA 4400, HP EVA CA, HP EVA CLX for Linux and HP ServiceGuard for Linux.

My customer has:

- 2x nodes Oracle 11g RAC is running on Red Hat Ent Linux 5 Update 2 at Primary Site
- 2x nodes Oracle 11g RAC is running on Red Hat Ent Linux 5 Update 2 at DR Site
- 2x HP EVA 4400, one for each Site
- 4x HP 8/8 SAN Switches, 2 for each Site
- 2x HP IP Distance Gateway (AG680A), one for each Site
- 1 T3 (2 Mbps dedicated inter-link between two sites)
- A lot of Cisco Routers & Cisco Catalysts for WAN connection
- The distance from Primary Site to DR site about 20km.

My solution diagram:

2 nodes RAC <--> EVA CLX + ServiceGuard for Linux <--> 2 nodes RAC
(EVA 4400 + SAN sw) (IP Distance Gateway + EVA CA) (EVA 4400 + SAN sw)

Could you tell me this configuration be going to work on fine, and it is supported from HP products & solution above?

I do not know how HP EVA Cluster Extension for Linux closely works with HP Serviceguard for Red Hat Linux for we can get DT objective.

Could you send me any reference docs for this case?

Moreover, I also knew HP StorageWorks Storage Mirroring 5.1 can help for this case with Host-based replication features. However, I am not sure the Oracle 11g RAC will be supported.

Could you give me your advises?

Much appreciate for your help!

Thanks and best Regards,
Nguyen Van Hai.
HP Viet Nam - hai@hp.com

[Ivan Ferreira]

Please consider reading the Oracle MAA architecture.

http://www.oracle.com/technology/deploy/availability/pdf/DRChoices_PPT.pdf

In our experiencie, combining multiple products to increase availability just increased complexity, manageability and support.

[Nguyen Van Hai]

Dear Ivan Ferreira,

Thanks for your information.

Regards,
Nguyen Van Hai.

[Steven E. Protter]

Shalom Nuguyen,


I do not know how HP EVA Cluster Extension for Linux closely works with HP Serviceguard for Red Hat Linux for we can get DT objective.

This hardware combination, my sources say is tested by HP.

Recommendations:

Have ilo updated on all systems and test fail over with a written test plan including expected results prior to deployment.

Have PSP installed on all systems and any utilities available for the EVA to insure maximum flexibility and monitoring.

Understand that what you are doing is building a fault tolerant system. Normally such a system will survive and run through one fault, not two.

You may get lucky on a second fault but don't count on it.

Overall a pretty well thought out plan.

SEP