EVA tender response

[Paul_M]

Hi experts,

I am working on a tender for a fairly large SAN deployment and I am putting forward an EVA 8100 as my preferred selection.

The customer has asked a question which I have seen raised once on this board before but with no conclusive answer.

They have asked if the EVA performs background "disk scrubbing" which amounts to background maintenance to ensure data integrity.

Does S.M.A.R.T disk technology cover this on the EVA or some aspect of levelling? Would appreciate any enlightenment.

Oh and one last thing , are there any general RAID rebuild stats kicking about. As in how long it takes the EVA to reconstruct a VRAID failure. This depends on many circumstances but has anyone come across any approximations or real life examples of VRAID failure rebuild. Are we talking minutes, hours?

Many thanks

Paul

[IBaltay]

Hi,

the drive polling
and data scrubbing
are implemented also to EVA storages:


Drive Polling :
On idle drives only, perform a Read of the EVA ID block area (every 10 seconds) to ensure it still has a valid stamp.

Data Scrubbing :
During idle time of a drive, the scrubbing process is designed to find media errors and correct them prior to the user reading the block.
(Background only. Cycles through all data (10 days or so)

[Paul_M]

Hi there,

Thank you very much for the answer. This sounds promising. Could you point me at the source for this information as I will need to quote this feature?
That would be really helpful and full solve my issue.

Thanks

Paul

[Peter Mattei]

Regarding rebuid times:
This very much depends on array load, occupancy, RAID level mix and number of disks in a disk group.

You can see from just several minutes on a lighly loaded EVA up to several hours on a busy and fully allocated EVA.

Rule of thumb:
The max recovery copy pace you can expect is about 10MB/s.
Assuming an idle EVA with a 300GB disk failing fully allocated with 270GB would result in a recovery time of about 7.5 hours.

An EVA with heavy IO can take several times longer!!

Cheers
Pete

[Paul_M]

Hi Pete,

Great answer, a rule of thumb will do nicely here, I am answering with best intention!

You wouldn't have any further info on disk scrubbing would you? Customer is concerned that an EMC rival bid is claiming this is a feature the EVA does NOT have. The initial response to my query seems to contradict this. Could you shed any more light on this?

Thanks

Paul

[Paul_M]

Hey IBaltay,

Any more luck in finding a reference to the disk scrubbing abilities the EVA runs?

Don't want EMC to get the drop on me here :-)

Will allocate points on this shortly.

Thanks

Paul

[Víctor Cespón]

The EVA controllers log every single incident on the internal loops, just ask anyone that has seen an EVA log in detail. If a packet is lost, arrives corrupted, a disk is not responding, etc it generates an ASC/ASCQ and it's logged.
Also, any trouble on the loops is logged and there are algorithms in the firmware to detect, identify and insolate any disk causing problems.
This works very well with the latest firmware on controllers and disks. If a disk starts having too many read or write errors, the EVA issues an warning, and if you have WEBES installed, it will e-mail you with the position and model of the disk that needs to be removed and replaced ASAP.