EVA3000 Purchase

[ROSS HANSON]

Hello,
My company is planning to purchase an EVA3000
Storage Device.

Does anyone have one and could you tell me if you have had any problems with it...whether it be firmware upgrades, replacing drives, controllers, updating any software...etc...

Thank you

[Uwe Zessin · Accepted solution]

Ross,
I have set up several EVA-3000 in the past. Four of them are running at a single customer site in a HP-UX environment (2 clusters, 2 datacenters). Except for a few disk failures which were covered by VRAID protection we did not see any problems.

I have also upgraded the earlier 2 systems from 2 to 4 disk enclosures on-site and everything went smooth. The last two eVAs I have upgraded before delivery to the customer.

The first two models also got a firmware upgrade from version 2 to V3.010 - again, everything went smooth. I have done firmware upgrades since version 1 on the EVA-5000 and I have never seen a problem with it.

[Mike Naime]

Unlike Uwe, I cannot paint a totaly rosy Picture for our EVA's.

We currently have four full EVA5000 systems. (240 drives each) Two were installed this last month. The other two we have had about a year.

Replacing a drive is a hot swappable deal that our local FE takes care of. (Part of the service agreement) We have been averaging between 1-4 drive failures a month (Out of 480 disks) This is higher than our HSG failure rate when you consider the 480:3000 or 1:6 spindle ratio. We have all 146GB disks in our EVA's. Our HSG's have everything from 9GB to 146GB drives.

We have not had any major problems with the EVA and our 200+ VMS sytems. Our problems have been mainly associated with the 24 AIX systems that use the SAN. One day we had an EVA drive failure, and it trashed disks on 2 different AIX clusters that where using that same EVA for their Oracle data drives. Fortunately these where not yet "Live" production systems. This occured just 2 days after applying all of the latest and greatest patches/kits on everything.

We also went through the 3.010 firmware upgrade. That was a 5 hour DOWNTIME for that storage system! :-( I call that lenght of downtime a "Problem". The upgrade itself went smoothly with no problems encountered. This downtime does not compare favorably to the 20 minute rolling upgrades, or the 5 minute "Shutdown" upgrades that we did for our HSG's. We have done other EVA rolling upgrades that did not require noticeable downtime. HP has assured us that we "most likely" will never have to take this kind of outage again. They would not guarantee that it would not occur again. (CYA on their part)

We always get HP field service to upgrade our SAN appliances/firmware, so I really do not know exactly what is involved there.

Adding/removing a large number of disks to an existing disk group can cause normalization to take several day to occur.

We have had "Ghost" drives. A disk failes. After being removed and re-placed. It still shows up on the system. Sometimes re-starting the Command View EVE process or re-booting the appliance can clear this up. We have had to reboot the EVA's to fix this in the past. (Pre 3.010 code)

Performance wise... It blows the HSG's out of the water.

Management of Space VS Managing spindles is also much easier.

The GUI tool is easy to use and understand. Like most other GUI's it can be dog slow at times.

My one big problem right now is accessing the SMA GUI while the backup SSSU scripts for 50+ production clusters are running at night. (about 25 terrabytes of backup data each night on the "Backup" EVA.) I waited one days for over 3 hours to get a GUI window that I could do something with! Fortunately, I was not dependant on using the GUI. You can still perform SSSU commands during this time.

We have not had a "shelf" failure yet. I have done three of those on HSG's in the past four years.

[Mike Naime]

Ross:

After reading my mini-novel, I wanted to add the following.

Murphy's law says that you are going to have problems with any storage controller that is available on the market. Electronics can and will fail eventually. You just hope that it is more later than sooner. (Roll the dice)

Overall, my EVA's have been less problematic than the HSG systems. The amount of raw storage that I have in 3 EVA 5000's equals the storage in all of my 36 HSG's!


Our next storage purchase will either be more 2C18D EVA 5000's or an XP1024.

Mike

[Uwe Zessin]

Oh, I didn't intend to sound too rosy, but I did not want to scare off, either.

I have once lost a whole EVA-5000's configuration with all data on it with firmware version 1 and we did have problems with the loops until we got loop switches and all I/O modules in the disk drive enclosures were replaced, but since then the system has been stable.

On the other hand I have never had to do with such complex environments that you do and I am not surprised that you hit some new wall from time to time.

With the EVA-3000 I really have not seen big problems so far, but that system is not so big like the 5000 - there is less to break ;-) And a high disk failure rate I have seen at one customer site which uses HSGs, too.

I fully agree with you that ANY storage system can fail - you only have to wait long enough. Sometimes we get asked by a customer if he can now get rid of his backup, because he has a RAID system, but that illusion can be destroyed easily ;-)