AIX DR Recovery

UNIX AIX
1

I will be going for a DR test next month and have few concerns regarding restoring user vg's

System which will be restored is on P750 (AIX 6.1) and the DR hdwr is P570.

I do have bootable mksysb of the system and also have savevg (with no data) using "savevg -r -f " for the user volume groups on the rootvg.

I am fine with restoring mksysb to recover the OS and rootvg.

My concern is regarding the user vg's. We have aroung 25 - 30 hdisks across 3 user vg's .

Question : Do i need to have the same number of hdisks and sizes to be allocated to the server at the DR site. ? or I need to modify the save vg data accordingly ?

2

If disk layout is different you can chose new hdisks from command line :

restvg [ -b Blocks ] [ -d FileName ][ -f Device ] [ -l ] [ -q ] [ -r ] [ -s ] [ -n ] [ -P PPsize ] [ DiskName ... ]

if not, system will try to use the same layout as in source lpar.

1 Like
3

Hi,

I am thinking a disaster recovery with safevg is a weird way to do it ... as you obviously need to save your data on a running system before the disaster occurs - and disasters happen not to knock on your door in advance to tell you that something is going to go wrong - only DR tests do.

What kind of storage do you use at all and don't you have any means to replicate your data to the DR site (like rsync, data replication on SAN level like SRDF or offside backups, GLVM or similar) that you could use instead? How much production data can your company afford to loose in case of a real disaster? And how much time will you have to recover - as installing a box completely from images takes a bunch of time you may not have on critical systems ?

Kind regards
zxmaus

4

We use DS8000 and DS4000 for our SAN . We do not have the capability to replicate data to DR site in real-time or to send data on a regualr basis. We had budget constraints and I believe we will work on the replication this year.

At the DR site , we request the offsite backup tape to be sent to DR location. Process at DR is to restore our TSM server first and after we restore the OS and uservg (defs) on the DR location restore user data from the TSM. For DR test, we get 48 hours.

---------- Post updated at 09:33 AM ---------- Previous update was at 09:31 AM ----------

Borek, Is this a standard practise to request the exact no of hdisks and size required for each LPAR prior to the DR test ?

5

If you are recovering everything from tape then it does make sense to restore everything to identically sized disks - but its not compulsory - you could just have one super goiant disk and restore everything to that. It would work - but there would possibly be a performance drop off.

What you didn;t mention in your description is how you restore the operating system mksysb? I'm assuming that you have a NIM server in your DR site for this purpose?

In which case the order would be:
Restore TSm Server
Restore mksysb to NIM server
Restore mksysb to LPAR
Recreate disk layout with savevg
Restore Data

It sounds like the crux of this will how and when will the storage be set up for the DR test? Do the Storage guys backup and restore the configuration you require or do they have to create it on the fly, or is it permanently allocated?

6

At this site we have bootable mksysb tapes for the servers. This would be a small DR test with only 3 servers.

My assumption is that the Storage and LPAR are already created at the DR iste. Looking at the DR documents it says the hardware is 570 with 6 lpars.

I will get these questions clarified thiis week during our meeting with the DR vendor.