Dell Dr4000 - Best Storage Optimization Settings?

[rborhara]

Any users backing up to a Dell DR4000 / DR4100 Storage Device. i seem to get timeouts during the merge process

There is only one setting i can see in the advanced settings for the backup job
Enable inline data deduplication (recommended)
Enable Exclude Swap file blocks from processing (recommended)
Compression level (none)
Storage Optimization - Local target

The above are from Dells Dr4x00 setup guide which is a few years old and i do have the latest bios version- in the past ive had to break down the backup jobs and remove a Vm or 2 from the backup job, but this is not always feasible

I have a Dedicated SAN switch 10GB iscsi, both the Equallogic SAN and Dr4x00 are plugged into these
The Veeam Backup Proxies are setup against each lun in the equallogic so no access issues - backup jobs utilise DirectSAN mode
I never seem to get timeouts during a Full backup but there arnt many of these i do

[foggy]

Rakesh, what kind of timeouts are you seeing? Your settings look optimal, however, keep in mind that deduplication devices are not considered to be best practice for the primary storage and fit more in the second tier in our reference architecture.

Also, the following thread might be useful.

[rborhara]

Hi the error message is
Full backup file merge failed Error: The specified network name is no longer available. Failed to flush file buffers. File: [\\%ipaddress%\MK-Backups-Live\DR4000-Daily-Backup LIVE Servers-11pm\DR4000-Daily-Backup LIVE Servers-11pm2015-10-03T230112.vbk]. Failed to backup text locally. Backup: [veeamfs:0:60b227b3-387f-4688-a9cd-349b6048b2dd (vm-35996)\summary.xml@\\%ipaddress\MK-Backups-Live\DR4000-Daily-Backup LIVE Servers-11pm\DR4000-Daily-Backup LIVE Servers-11pm2015-10-03T230112.vbkVBK: 'veeamfs:0:60b227b3

yesterday the merge completed on the 2nd retry same for the night before
Dell have conformed the Dr4000 doesnt time
Veeam say its down to the way CIFS works with windows machines, i dont get why i see this only on merges and not backups in that case.
I do have the cleaner without a schedule which means it runs as required. Replication jobs dont run when the above backup job is running

[foggy]

Am I understanding right that you've already contacted Veeam technical support? What is your case ID?

[rborhara]

Hi foggy issue have been loggeda few times , breaking the backup jobs with less VM's seems to make a difference then it seems to haunt me after a few weeks when the merge issue reappears

00980402
00936116
00961284
00953051

[foggy]

As far as I can see, these cases are closed. If you continue to get errors, I'd recommend opening a new one for closer investigation.

[ChuckS42]

Any users backing up to a Dell DR4000 / DR4100 Storage Device. i seem to get timeouts during the merge process

I strongly recommend you open a case with Dell for this. They have a whole bunch of tweaks you can perform that will help smooth things out in general. I use a pair of DR6000 devices, and had a ton of issues similar to yours; they were able to help quite a bit. Ultimately Veeam isn't going to be able to help much on this subject (unless they are working directly with Dell), since the issues you are seeing are with the storage device.

Unfortunately, however, you will not be able to run anything resembling a synthetic backup (including Forever Forward Incrementals) on the DR-series devices. The merge process plays holy hell with the storage box and gives you the outcome you are seeing. These things are deduplicating appliances; when you do a merge it has to fully re-hydrate (i.e., re-duplicate and uncompress) both the Full backup and the Incremental you are merging, while at the same time de-hydrate (dedupe and compress) the output. The DR devices cannot do this well at all -- performance goes to crap very fast.

Your only option for the time being is to switch to a traditional Daily Incremental/Weekly Full model. Disappointing, but it works far better.

I hope that someday Dell and Veeam can work together to sort these issues out, because otherwise the DR-series boxes are very nice.

[ChuckS42]

Oh, and for what it's worth, these are the basic storage settings you should have in your backup jobs:
Backup - Incremental (no synthetic), with a periodic Active Full (weekly)
Storage - Enable inline data deduplication selected; Exclude swap files blocks selected; Compression level set to None; Storage Optimization set to Local Target.

It looks like you've got the right Storage settings.

[mongie]

unless they are working directly with Dell

I heard something to this effect... RDA support would be a miracle. Pretty much the only thing that is going to make these appliances work nicely with Veeam.

We use DR4100's too, and the truth is, if you have substantial backups (size and retention), there really isn't much you can do to get around the merge and transform times (I'm talking about backup scheduling, veeam settings etc.) Unfortunately, I've come to realise that even though using faster storage as 1st tier for backups helps... the truth is, you still have transforms and health checks if you use DR4100 as a backup copy target. I'm interested in how the active backup copy feature will work with v9, but ingest is still a problem... and instead of spending 24-48 hrs doing a transform of my 3TB Exchange server job, i'll be spending 18-24hrs copying a full backup from primary to secondary backup storage.

My advice to ANYONE thinking about buying one... Think again. Look for an appliance with API support (StorOnce, DataDomain), and even then, consider plain disk - at least you know how much storage you're going to get. Hell, once Dell has bought EMC, I highly doubt the DR series are going to exist.

[rborhara]

Hi guys thanks for the updates, i have spoken to Dell and its a case of playing cat and mouse between Dell and Veeam, maybe I have to resort to the fact the Dell Dr4000's have reached their limitations

[JMP]

We are also using DR appliances and the only real problem is backup copy jobs. It's just not possible to use backup copy jobs because merge operations are too IO heavy.

We are now taking backup directly to appliances and performance is good enough, ingest is 400-600MB/s per appliance and that is enough for us (we have about 80TB of VMs). We are waiting for V9 and hopefully after that it is possible to use Veeam's "Ultimate architecture" with deduplication appliances...