topic Re: Planning Disaster Recovery-Technical Issue in Operating System - HP-UX

Planning Disaster Recovery-Technical Issue

Misbakhul Munir — Mon, 08 May 2006 06:17:05 GMT

Hail all,
Our company plan to make disaster recovery system. The goal is to make rocovery in minutes when problem is accured. Our DR site about 300 KM from Main site and maximum Bandwidth available is about 2 Mbps.

this is current our machine:
1. 2 HP rp5470 with mc/s service guard
2. 2 HP DS2405 storage
3. Running HP-UX 11 only (not i version! )
4. Running specific apps in filesystem (vxfs)and ONLY run on HP-UX 11 (not i version).

According with our situation, what software (and or hardware) we must purchase ?. FYI, we have another 2 rp5470 and 1 VA 7110, and we plan to use that hardware.

Regards,

Munir

Re: Planning Disaster Recovery-Technical Issue

melvyn burnard — Mon, 08 May 2006 06:28:45 GMT

Perhaps this document might help you:
http://docs.hp.com/en/B7660-90017/B7660-90017.pdf
Also, from what I understand of your question, you are running HP-UX 11.00, an dthis OS version goes out of support at the end of this year, you should plan to update to at least HP-UX 11.11

Re: Planning Disaster Recovery-Technical Issue

Steven E. Protter — Mon, 08 May 2006 06:51:44 GMT

Shalom,

If you want recovery in a few minutes, you may want the Metro cluster product from HP. This will fail over your applications to the second cluster. That is if the product is offered on 11.00 HP-UX, which is close to its de-support date.

You may wish to use a product like Oracle dataguard, or parellel server if you have any oracle databases. Parellel server can keep the databases at the DR site up to date.

I think your biggest problem is an OS thats about to go out of support. I'd do this.

Build the new rp5470 servers at your current data center with 11i v1 or v2. Migrrate your applicaition to the new systems.

Make Ignite backup images of the two rp5470 servers and restore them to the old servers, which are now the DR servers.

Interupt the Ignite restore and change the hostname and networking information of the DR servers.

Ship the old servers to the DR center and bring them online as hot DR systems.

SEP

Re: Planning Disaster Recovery-Technical Issue

Misbakhul Munir — Mon, 08 May 2006 07:03:36 GMT

As i mentioned before, it's most impossible to upgrade the O/S to 11i, because the application that run in this server.

when it's possible to built metrocluster or metrocontinental, how much b/w we must provide, and is there data always update (always sync each other) ?

is there any solution from 3rd party vendor for my case ?

Thanks

Re: Planning Disaster Recovery-Technical Issue

Greg Vaidman — Mon, 08 May 2006 15:03:08 GMT

metrocluster supports data synchronization via either Continuous Access XP (for HP's XP disk arrays) or SRDF (for EMC disk arrays). If not using this hardware, you'll have to figure out another way to sync data (rsync, DataGuard, etc.)

you'll have to be cautious with any tool outside of metrocluster used to sync filesystem data, because your cluster contains binary configuration data as well as configuration files with hardcoded IP addresses and such, so syncing them in full to a DR site that is live on the same network won't work - you'll have to manage the cluster config files separately. You wouldn't want to start your DR cluster and have it blow up your production cluster!

Re: Planning Disaster Recovery-Technical Issue

Misbakhul Munir — Mon, 08 May 2006 20:11:48 GMT

According to Greg's opinion,
if we use rsync or dataguard, where i can found this solution for DR ? and it's running well for mc/s cluster integration ?

Please help me..
Any solution that not necessasary our current O/S version and current production system will applied to our system

Thanks

Re: Planning Disaster Recovery-Technical Issue

Prashant Zanwar_4 — Mon, 08 May 2006 20:49:16 GMT

It would be good to consider EMC SRDF option for DR. This helps you keep your data upto date on remote site almost accurate and not much of lag..
There can be scripts written based on your requirement to sync application code-trees.. else again r1/bcv/r2 EMC stuff can help you do this..
If you are planning to take build system OS also when disaster occurs, ignite is option with taking images to DR site with SRDF some similar technogy..
Good luk
Prashant

Re: Planning Disaster Recovery-Technical Issue

Misbakhul Munir — Mon, 08 May 2006 22:09:59 GMT

Prashant,

i think SRDF is overkill for our system
is there another solution that with inexpensive and meet with our requirement ? or opensource solution ?

Thanks

Re: Planning Disaster Recovery-Technical Issue

Rainer von Bongartz — Tue, 09 May 2006 01:44:58 GMT

You might want to have a look at this product:

http://www.kashya.com/pdf/Datasheet_KBX5000_CRR.pdf

Regards
Rainer

Re: Planning Disaster Recovery-Technical Issue

Misbakhul Munir — Wed, 10 May 2006 07:46:31 GMT

Thanks for every body for your suggestion.

i think we must consider about the O/S version and storage version thats we use now.

Regards,

Munir