VA & EVA question

[John Jimenez]

This is a 3 part question.
Snap shots on VA7410 & 2 LUNS on 1 volume group

1) I just removed the windows LUNS off our VA7410 (59 disks)and now have freed up some space, so we now have room to add back in the snapshot of the 100 gig production environment. Right now we are utilizing only 2 of the 4 HP Tachyon XL2 Fibre Channel Mass Storage Adapters, and therefore have 2 redundancy groups.
When creating a snap shot is it more efficient to snapshot to the same redundancy group or to the other one? Or does it make a difference?

2) For ease of identification (LUN 108 (even number) = redundancy 2 = /dev/vg08, LUN 109 (odd number) = redundancy 1, /dev/vg09, etc…) and to allow the VA autoraid to control optimization each volume group only has 1 virtual LUN (1 seperate logical disk, but all use the same physical 59 disks). But our data is increasing rapidly so I often need to increase space. Which means I need to destroy and recreate the LUNS to a larger size. This is not big deal with the Development and Testing Environments, but is inconvenient with the production environments. I was thinking of utilizing performance by making 2 equal LUNS in both redundancy groups for each Production environment. True it would use the same physical disks, but would split the paths and load balance. But then I second guessed myself and realized that it might be counter productive for the following reason. The HP/UX 11i operating system does not load balance between these Primary and Alternate Paths. So it will use the LUN on the 2nd Redundancy Group inefficiently, in fact it will be slower. Is my theory right or wrong? And if I create a 2nd LUN of the same volume group into the same Primary redundancy group it will still use the same 59 physical disks and same path as the first one-- so nothing seems gained. Also, My database does not have the ability to quiese, so if the volume group is split up on different LUNS will Snapshots on 2 smaller LUNS take longer then 1 larger one making it less reliable?

3) Here is one last question, on the EVA (not the VA). I understand that the windows OS can utilize increased size when a LUN on the EVA increases, but will the HP/UX 11.11 recognize it? IF not there seems to be no advantage in a HP/UX environment. Are any of answers to the 2 questions above differnt on the EVA as opposed to the VA?

Thank again,
John

[RAC_1]

John

Very long queston.

1. What do you use for snap shot?? Online JFS?? IF yes, you just create a lvol in any of the redudancy group and mounts a snap on this lvol. so does not matter much. In any case, for some luns (lvols on hpus)you will be mounting then from same redudancy groups.

2. The use of alternate pv paths is for failovers and not load balancing. (You should be using secure/auto path for that-additional $$ product) But what you do is as follows.

for vgx and lvolx in that vg, keep controller a as primary and b as secondary. Now for vgy and lvoly in that vg, keep b as primary and a as secondary. This would ensure your i/o go though both controllers. Again, you can use two luns in vg and make different primary and alternate paths to optimise the the use of controllers.

3. On hp-ux, if you increase the LUN, size you will have to do required changes on hp-ux side. If you have online jfs, it is easy and can be done on the fly.

extend the vg, under which that lvol lies.
vgextend /dev/vgxx /dev/dsk/cxtxdx
lvextend -L "new_size_in_MB" /dev/vgxx/lvolx
fsadm -b "new_size_in_MB" /mount_point
Check with bdf.

If you do not have online jfs, lvextend, unmount and extendfs

Anil

[John Jimenez]

RAC,
Thank you for the response

1) I am using the VA via Command View to do the snapshots. But it sounds like it should be the same and not matter.

2) I forgot to mention, I have only 1 logical volumes in all my volume groups. I have a 2000GB LUN Security License on the VA, But I am pretty sure this is not it. I will look into this load balancing software on HP/UX. If you already know a link that would be super.

3) So they finally fixed it. About 6 months ago I saw a demo and even though the LUN on the EVA was increased, the I/O Scan of HP UX was stil reporting it as the smaller size. Glad to see that it works now. Maybe now is the time to upgrade to the EVA.

3)

[Tim D Fulford]

On point 3... IF you are using LVM to manage your logical volumes, then you CANNOT use on-line JFS, as the physical volume (vDisk) is fixed in the LVM header... that said, if you put your filesystems straight onto the evaxxxx vDisks, then you can... as most people do the former then you cant..

Regards

Tim

[John Jimenez]

Tim,
Right now the RP7410 is using on-line JFS. If we upgrade I would plan on using it again. But do I understand you that it only recognizes it when you "do not" use on-line JFS