Disaster Recovery System

[Former Member]

Hi all,

We are SAP 4.7E , HP UX 11.11 , Oracle 9.2 .HP UX mirrored configuration RAID 1. We are planning for disaster recovery system for our present configuration.So I would like to know what would be the Hardware and Software requirements for this.I request your valuable suggestions on this matter

Thanks and Regards

Varughese

[former_member204746 · Accepted solution]

it all depends on what disaster recovery solution you want.

a few possibilities:

1. install your QAS server in another city and use it as a DRP server in case or PRD issues. Calculate around 24 hours downtime for this strategy. no hardware costs.

2. create a DRP instance in another city and ship database logs on this other server and apply them when you receive them. this costs additional hardware. downtime is less than 1 hour.

3. High availability with cluster, this costs a lot in hardware and software, but downtime can be virtually eliminated.

so, what kind of strategy are you looking at?

[Former Member]

Hi!

Last year I've been part of a Metrocluster project with HP in a customer running several SAP flavors, and it works very well.

The only problem might be the cost. In this case our customer is using active/passive sites, but you can use active/active sites to have a better ROI. Of course... you need to have a good network to use in it active/active.

http://h71028.www7.hp.com/enterprise/downloads/5982-9585EN_SG%20manageability_WP_final%20review.pdf

Cheers,

FF

[Former Member]

Hi

I am working on DR for SAP with oracle databse 9.2 if any one has done the Disaster Recovery please help me.

Regards

Vishal

[Former Member]

Our aim is to keep the data safe rather than reducing the down time for example a system in another building in the same location.

Can we set up a third copy of the data in another location.What would be the hardware requirements for this?

Thanks,

Varughese

[former_member204746]

Ok, then check this:

setup a server that is instlaled offsite, setup a network link between 2 sites and ship datbase logs over the network... then apply redologs on the standby database in other site. Chec kORacle DataGuard and DB Shadow for third party solutions.

make sure that you take a copy of your backup and keep it offsite! People tend to forget to keep their backup as far away from their servers!

[manfred_stein]

If you want to achieve this on storage level, 2 SAN or NAS storages not too far from each other connected over FibreChannel or GB Ethernet which are doing the synchronous mirroring (or go with just the RAID protection, if this is sufficient), in addition a not too slow network link (100MBit might do) to a third location where you mirror the data asynchronously.

The nice thing is that this can cover all data pertaining to the system.

[Former Member]

2 storage units as far away as possible where the data gets cross replicated, make sure backups are in good healthy state and ship a copy to sungard , make sure the nfs mounts are taken care of .

[manfred_stein]

Complementing Eric's answer, there are several degrees of protection against disaster.

1) Synchronous mirroring of data to 2 (or more) storage devices, e.g. 2 or more SAN storages. All recent Unix and Linux releases support this using host-based mirroring technology. SAN storage or NAS will very often come up with additional feature packs providing the same functionality on storage level. Works up to around a mile of distance between the storages, latencies become too big if distances are larger. In Linux you can search for MD Multiple Devices, Red Hat's CLVM since Rhel4.4 has been extended with loads of functionality as well.)

2) For larger distances (e.g. building up Metro Clusters) very often asynchronous technologies are used to mirror the data to a second location. Again this can happen on OS or storage level. In Linux DRBD is a good search term.

3) There are also continental clusters, in some decades planetary clusters may become available:-)