questions about EVA 4100 vs CX 3-10 and others

[Patrice Bourdon]

Hi, i have to chose between adding an EVA 4100 with continuous access replica on my EVA 3000, and i have some questions about this, experiement return appreciated if you have it...
- what about perfs EVA 4100 (or 3000) vs EMC CX 3-10 ?
- what about EMC management console vs Command view 7 ?
- should i really change Disk failure protection from "single level" to "double level" and loss a big amount of disk space when i install CA ?

Thanks for help and sorry for cross-posting (disk array group and san group...)

[IBaltay]

Hi,
I am sending you the link on the EVA 4400 and CX3-10 comparison marketing video
http://hp.feedroom.com/index.jsp?fr_story=f8a49df7aba9478a746a7c57075f473ff40e3188&fr_chl=d9138bf1d80fad18e3bfa58c2dc62ae5716c10df

[Patrice Bourdon]

thanks for the link, even if it's the "HP point of view"... i think that someone from EMC will send me a link to say that CX is better...
But i already compared EVA 3000 and CX 300 4 years ago, and found that HP was best (and bought an EVA), so i don't think things have so change !

[Patrice Bourdon]

thanks for the link, even if it's the "HP point of view"... i think that someone from EMC will send me a link to say that CX is better...
But i already compared EVA 3000 and CX 300 4 years ago, and found that HP was best (and bought an EVA), so i don't think things have so change !
what about double leveling for continuous access ?

[Sheldon Smith]

Disk failure protection (none, single, double) simply changes the amount of total raw storage available for provisioning. When a spindle fails in a disk group, the EVA always uses free space within the disk group to rebuild the RAID5 and RAID1 blocks as needed. Disk failure protection is a safety net to keep you, the storage administrator, from allocating too much raw storage and thereby leaving the EVA in a state where it could be unable to recreate the lost information at the time of a failure.

Single tweaks the total raw storage available to hide enough space to recreate the largest spindle in the disk group assuming it was full and all VDISKs on it were using VRAID1; hence two physical spindles-worth of space. Double hides enough space to recreate *two* disks, therefore hiding four spindles-worth.

The actual question is: How close is anyone monitoring the health of the target EVA? If "close enough", then Single would be fine, as people will notice the failed drive and replace it before a second drive fails.

Hope this helps.
By the way, don't forget to assign points (0-10) to ALL of your responses. At the moment, you have assigned points to only 5 of 16 responses.

[IBaltay · Accepted solution]

Hi, if your double-levelling means double protection, that precisely means you have a spare space of 4 disks, in a single protection method setting it is a spare space of 2 disks. This protection lays on the disk group level, so you will loose the capacity of 2 (single protection) or 4 (double protection) harddisks. It has nothing to do with the vdisks which are being replicated.

[Patrice Bourdon]

Sheldon, my question about leveling was relative to Continuous Access, one of my vendors said that with Continuous Access chage leveling from single to double is mandatory and so on i'll loose half of my disk space... it seems strange, no ?

ps : where can you (and i...) see the answers not "pointed" ?

[Sheldon Smith]

Very strange. I do not know what the vendor meant by "double leveling". We are assuming he/she meant the disk group protection level. Setting None will show you all your raw storage space, Single will reduce it by twice the largest spindle in the group, and setting Double will reduce the total capacity by a total of four or the largest spindle in the group. And yes, you can have a disk group consisting of disks of different size.

You should be able to click on your name link. When your ITRC profile comes up, scroll down to "My questions". There is a link there to "Questions or topics with unassigned points".

[Uwe Zessin]

Hm, interesting. I've read a lot of the CA documents and have never seen a requirement to use 'double failure protection' in CA environments - can you please ask your vendor which document says so?

Double protection is really meant for EVAs that are not under constant monitoring for a long time to be able to cover more than one disk drive failure.

Since some years, however, it is fairly standard that EVAs are monitored and faults are automatically sent to a HP service center.


Now, if all links between both EVAs break, all write are recorded in the write history log (WHL) by default. This takes space, but the EVA does not use the space reserved by the 'protection level' for the WHL so it still does not make sense to use 'double'.


> and so on i'll loose half of my disk space...

Huh? You disk group contains only 8 disk drives? Unless you are running with mixed sized disks, that's the only way you can lose half the space (or my math is flawed ;-)

[Patrice Bourdon]

hi, my actual disk groups are :
- 8 X 72 GB FC
- 12 X 146 GB FC
- 8 X 500 GB FATA

[Patrice Bourdon]

and my EVA has ISEE active and i monitor with CV 7, so i think double leveling is not necessary, i'll ask the vendor why he said that.

[Rob Leadbeater]

Hi,

I recently had to setup a CX3-20 and with respect to the management interfaces, Command View EVA wins hands down in my opinion.

OK, you don't need an additional server to run Navisphere, but the process of getting things setup is just tedious if you're used to Command View.

What should be simple tasks like adding a HBA port WWN to a server, seems to send you round in circles in Navisphere... No doubt someone who is used to the EMC product will contradict me here, but I just think its overcomplicated for what it does...

Cheers,

Rob

[IBaltay]

Hi Patrice,
your disk groups are not very optimal from the performance point of view. According to the best practices the more disks/spindles in the disk group, the more performance you can have on the vdisks

[IBaltay]

Hi,
for the future I suggest to extend only one FC DG (e.g. 146GB) to achieve better performance

[Patrice Bourdon]

Thanks for your feedback Rob, it confirms what i saw 3 years ago...

IBaltay, the reason why my diskgroups are not optimal is that i begun with 8 X 72 GB, then populate with others, i had no budget for buy 24 FC 146 GB. I'm buying a new SAN (EVA, i guess), and again i buy 8 FC and 8 FATA, next year i'll upgrade the number of each group (if i get money... #;) )