Comprehensive data protection for all workloads
Post Reply
yowmemperor
Enthusiast
Posts: 30
Liked: never
Joined: Jan 08, 2018 5:19 pm
Contact:

Storage Speed and Protocols

Post by yowmemperor »

We currently have 9.5 U3, and have a copy job to a remote site on a Dell DR4100. Speeds are slow, the DR4100 is the bottleneck, reaches up to 120mbs max. This makes health checks, synthetic fulls, and tape outs painful.

I was advised to switch to OST rather than CIFS for a protocol and it should triple my speed. How is best to go about this? Create a new container and copy all the data in the existing one over? I noticed I can't change CIFS to anything but NFS. We have 30TB of files in the current directory so I am concerned about the ability to copy that much data to a new folder. In addition, its a 50TB device.

Any advice is welcome!

Thanks,
Matt
PTide
Product Manager
Posts: 6408
Liked: 724 times
Joined: May 19, 2015 1:46 pm
Contact:

Re: Storage Speed and Protocols

Post by PTide »

Hi,

May I ask you who gave you that advice? The problem is, that Veeam does not support OST. OST was created to enable integration between disk-based storage appliances (such as Dell DR4000) and Symantec’s NetBackup (version 6.5 and higher) and BackupExec (version 2010 and higher). Unless OST stands for something different, it won't work with Veeam. Another question - do you have gateway server deployed on the remote site?

Thanks
yowmemperor
Enthusiast
Posts: 30
Liked: never
Joined: Jan 08, 2018 5:19 pm
Contact:

Re: Storage Speed and Protocols

Post by yowmemperor »

The advice came from Quest I believe. I saw some of that info as well, just figured maybe Veeam supported Symantec protocol after they said that. They also said to configure it to use multiple streams under CIFS for better speed. Not sure how to do that. We do not have a gateway server deployed. That article ill have to read again in the morning. I am a bit lost.

We have two proxies, the one on the remote site copies backup files to the 4100, no WAN acceleration needed. The local site copies directly as it is the next rack over. The tape device is two racks over from the 4100 connected to the tape server via iSCSI. Is this saying I should have a gateway between the tape server and the DR4100? It also says it will use the tape server as a Gateway, but then again, I don't fully understand the article. (It's the end of a long day).

I appreciate the info!
mongie
Expert
Posts: 152
Liked: 24 times
Joined: May 16, 2011 4:00 am
Full Name: Alex Macaronis
Location: Brisbane, Australia
Contact:

Re: Storage Speed and Protocols

Post by mongie »

Fellow DR4100 user here. As of the most recent firmware, there is support for RapidCIFS under Veeam. As I understand it, RapidCIFS essentially only sends "new" blocks across the wire to the DR4100, saving bandwidth. With traditional CIFS you'd be sending blocks across the wire to the storage only for it to be discarded due to already being on the disk.

You have two main options to optimise performance.

a) Use Quest RAPID CIFS (which is a small driver that gets installed on the Veeam proxy / storage gateway on the source side). This method tries to optimise perfomance by using the DR4100's intelligence to minimise traffic going across the wire. The downside is that what you're copying across the wire needs to be uncompressed.

b) Use Veeam compression to a gateway server or remote proxy. This way you configure Veeam to compress the backup (as much as possible) and transfer compressed data across the wan to the Gateway server or remote proxy, where it is decompressed and written to storage. The downside here is that if you already have a "full" backup on the DR4100, you're potentially transferring "known" blocks across the wire.

You could definately use Rapid CIFS with option b, but I think it would have a minimal impact. Personally, I'd go for option A.
PTide
Product Manager
Posts: 6408
Liked: 724 times
Joined: May 19, 2015 1:46 pm
Contact:

Re: Storage Speed and Protocols

Post by PTide »

I don't fully understand the article. (It's the end of a long day).
I should have provided you with this one, perhaps.

In VBR, when you setup a CIFS share-based repository, you are provided with an option to specify a gateway server to be used to host Veeam data mover service. In your case you can pick any Windows server on the remote site to serve as a gateway between your local repo and remote DR4100. As Alex has correctly described, with a gateway server present on the target site backup traffic can be reduced by Veeam before it goes into WAN. The difference between two options is who will do all that dedupe and compression heavy lifting - Veeam or Quest.

Thanks
Post Reply

Who is online

Users browsing this forum: Bing [Bot] and 248 guests