EVA 4400

 
Derar_1
Occasional Advisor

EVA 4400

ON EVA 4400, In case out of 8disks with each Enc that is 16 hard drive at maximum.How many hard disk failure will damage the data ?



7 REPLIES 7
Sivaramakrishna_1
Honored Contributor

Re: EVA 4400

hi,

If you have 16 HDD then your RSS count will be 2 (there will be 8 HDD for each RSS generally).

In such case, if you have any vraid5 volumes, the maximum of 2 Disk failure at a time can be handled subjected to one disk failure in each RSS.

If it is vraid1 maximum of 8 Disk failure can sustain the data - 4 HDD failure in each HDD and those HDD should not be part of a mirror pair in that RSS, more over this will be possible if the Disk Group is "Basic".

In case of "Enhanced" maxium 2 Disk Failure in RSS can sustain in case of vraid6 or vraid1 even.
Johan Guldmyr
Honored Contributor

Re: EVA 4400

Or if you are running vraid0 it just takes one disk to pooch the data ;)
Prakash Singh_1
Esteemed Contributor

Re: EVA 4400

Hi,

For EVA4400 If you have one disk group with 16 drives you would have 2 Redundant Storage sets.

At a disk group level you can set your failure protection level to single or double drive failure. This means that your disk group can sustain upto 1 or 2 drives failure based on the protection level you select.

Apart from this at vdisk level If you have 16 drives then your DG would have 2 RSS With drives in each RSS.

For vraid5 volumes, 2 Disk failure at a time can be sustained if there is one disk failure in each RSS.

For vraid1 volume maximum of 8 Disk failure can be sustained the data.

For "Enhanced" maxium 2 Disk Failure in RSS can sustain in case of vraid6.

Regards,

Prakash
Regards,

PS
To assign points on this post? Click the white Thumbs up below!
Mark...
Honored Contributor

Re: EVA 4400

Hi,
I think a couple of people need to be *very* careful with your answers.
My understanding from the question is that there is an EVA with TWO disk encl with 8 disks in each encl.
Stating that the maximum amount of disks that can be lost gives the impression that you can lose up to that maximum and still maintain data! If any Vraid0 disks exist they will be lost with one disk failure. Even with Vraid1, ok lose one disk but what if the next disk you lose is the mirrored pair of the first disk ?? and it fails whilst he first disk is rebuilding ???
If a disk fails then it should be replaced as soon as practically possible and an EVA with two disk encl is not the best configuration for maximum resiliance of the product.
Mark...
if you have nothing useful to say, say nothing...
gregersenj
Honored Contributor

Re: EVA 4400

True Mark.

vcespon wrote a very good comment on this some time ago.

Don't try to calculate. how many disks you can loose.

Do try to understand how stuff works.
Any RAID is at risk catastraphic disk failure.

First undersatn the risk, with traditional RAID.
RAID 0. 1 disk failed = restore from backup.
RAID 1, with hot spare.
If a disk fail, it will have to recontruct to the spare disk. Within the recunstruction period, the RAID is at risk, untill reconstruction has completet.

RAID 5, with hot spare, you are at risk for addtional failure untill reconstruction has completet.

RAID 6, you can loose 2 disks.

RAID 1+0, with hotspare.
RAID 1+0 is a number of mirror sets that is striped. If a disk fail. your are at risk on the mirror set that got a failed disk. If the other one fail, then the RAID is down. But you can loose a disk in the other mirror sets.
Here it all depend on how multible disk failure hit you.

The EVA:
VRAID 0 is RAID 0 - same rules.
VRAID 1 is RAID 1+0. Same rules.
VRAID 5 is RAID 5+0. Same rules as a combination of RAID 5 and RAID 1+0.
VRAID 6 is RAID 6+0.

The RSS is an optimation of the security.
I can't tell much aboute that.

When an EVA get a disk failure.
It copies/reconstruct/migtate the data, to the space that has been reserved for disk protection.
After that it will do a new levelling.
So it all depend on how unlucky you might be.

BR
/jag

Accept or Kudo

Derar_1
Occasional Advisor

Re: EVA 4400

Hello All,
Thanks All for your Reply...

Now I have one disk group(Default Disk group) with 16 HDD with 2 RSS ,single drive failure.

Case 1:-
and I have one Vdisk Mailbox with 750GB vraid1 ,when 9 HDD failure in this Default Disk Gruop then i will loose my data ?

Case 2:-
Now if i have two Disk groups each with 8 HDD,and I have one Vdisk Mailbox with 750GB vraid1,when 5 HDD failure in the Default Fisrt Disk Group then i will loose my data ?

And if and I have other Vdisk SQL-DB with 500GB vraid5 in the Second Disk group , when 4 HDD failure in this Disk Group then i will loose my data ?

And what is the best practice for this Case is to have 1 Disk Group with 16 HDD or to have 2 Disk Group one for Exachane Servers ,the Second for the MSSQL ????
gregersenj
Honored Contributor

Re: EVA 4400

No you can't put it like that.

If you get 9 disk failing at the same time, you most likely will loose data.

In case of disk failure. the EVA recontruct the missing disk, then it do a levelling.
Wich mean it redistribute all data across the disks.
Then you are back at the same security as before the disk failure. But you got less free space.

If free space get to low, the EVA will choke on it.

I will recommend 1 disk groupe of 16 disks.
1. Higher performance.
2. Lower overhead for Disk protection.
3. You can't ungroupe a disk, if you only got 8 disks.

You create mutible VDisks within the same disk groupe.

Ei.
You got 16 disks.
Create 1 diskgroupe.

Lets say you need 750 GB for SQL.
You greate a 750 GB Vdisk. RAID level can be 0, 1, 5 or 6. You decide the RAID level when you create the Vdisk.
Then you need another 750 GB for exchange.
You create a second Vdisk of 750 GB, and choose the RAID level that you wish.

VDisk's within the same disk groupe, can be of various RAID levels.

You can add more disks to a disk groupe later - on-line.
You change the size of a Vidsk on-line.

Just remeber, always to have sufficient free space on the diskgroupe.

BR
/jag

Accept or Kudo