Flare Disk's rebuild problems

There has never been a recommendation not to use the Vault drives for user data.

Failures to Vault drives will always schedule a rebuild when replaced.

With regard to Vault drive usage, it is recommended that you be careful with provisioning RAID groups containing user data on the Vault drives.  It is possible for RAID groups created on the Vault drives servicing a production workload to impair overall system performance, or adversely affect storage system availability.

The specific recommendation including usage recommendations for the CLARiiON CX4 Vault drives can be found in the EMC CLARiiON Best Practices for Performance and Availablity, FLARE Revision 29.0 in the Vault drives section.  Please note the  'Maximum host I/O loads for Vault drives' table.  A similar section exists in the earlier EMC CLARiiON Best Practices for Fibre Channel Storage, FLARE Revision 26.0, should you be using a CX3 or earlier CLARiiON model.

With regard to Vault drive rebuild, the information on how it is performed can also be found in Best Practices in the 'Vault and PSM effects' section.  In addition, the 'Vault and write cache availability', section in EMC CLARiiON Storage System Fundamentals should be reviewed. This document is written for users new to the CLARiiON CX4-series.  Pay particular attention to the example at the end of the section.

Both the BP and Fundamentals documents are available on Powerlink. It is always recommended that you reference these documents ahead of 'rumors'.

Thanks JPS00 (why hiding ?)

I'm will not quiet agreded whit you statement "There has never been a recommendation not to use the Vault drives for user data."

No mabyee NOT an official statement, but as good latin as you newer know what I/O load you users put on the LUN.

On the Clariion Performance Course the EMC Instructor Steven Stead, stated to never put user data LUN's on Flare drive for best array performance and stabality, and also on the CX4 Installation course in Brentford UK, its also stated not to use these drives.

Also here at StorageNerve they recommend NOT to use disk 0_0 - 0_4 Flare disks for user data !

http://storagenerve.com/2009/01/06/emc-clariion-flare-code-operating-environment/

Here i my company, we have expirenced that the array went in a non responsive state, becourse an HI I/O loadet LUN on the flare dives, bring the flars in knees, and the support recommendet to move (migrate) the lun to anohter area.

So in my point of wiev, a proffesional storage administrator, don't dare to take the challange to bring our array to the limit !

As stated in the Maximum host I/O loads for Vault drives' table 8 in CLARiiON Best Practices for Performance and Availablity, FLARE Revision 28.5

Page 37 you can use 100 I/O's whit 10 MB/s bandwidth, but it will affect Navisphere manager commands to respond slow (delayed), but are we shure that we don't over load the drives ? No im not shure !

And whit these cheap drive prices we have today, we can afford to use 5 dedicated drives for flare, for best array performance and stability.

So i have 4 Clarrions whit no RG/Lun's on the flare disk's

Just my 5 cents

Back to the "Rumor"

I have heard from the Dell/EMC Support that there have been some ocations where the hot spare not have kicked in, if a flare drive was faulted, and there not have been any user RG/LUN on flare drives.

Any one expirenced this ?

This can be dangerous for us, who don't dare to have user data on flare drives, therefor the dummy lun's i talked about.

And EMC instructor Russel Starling (EMC Milford US) has talked about this issue, but cannot find him, so i can ask him into this, by mail.

Regards Kenneth Hansen