Veeam Distributed Deployment

[ledvanb]

Dear Colleagues,


im planning to deploy an Enterprise 3 site distributed backup managing it with Enterprise Backup Manager, with Veeam + HP storeonce + HP tape libraries.Each site will have:

1. One physical DL380 G10 with 256G RAM + 16GB FC and 10GB Ethernet
2. Storeonce 5660 with over 1PB capacity + tape library

Each site will have its own backup schedules as well as replicating its backup to another site through storeonce Catalyst Copy in a ring architecture,im planning to use the single physical server as backup+proxy+gateway server to utilize Direct SAN transport Mode in an efficient Manner( LAN fee backup).At the moment i will not use a Second sever on each site but i might in future,there are some points i would like to confirm,so i appreciate any and all recommendations/comments:

1. Is there going to be an issue running all roles on a single physical server(backup+proxy+gateway)?

2. While the Storeonce appliances are replicating through catalyst copy, will the proxy(data-movers) & gateway servers roles be involved? considering the deduplication is done by storeonce? the reason im asking this is that i want to know if i should consider any extra load on our physical server on each site whenever the catalyst copy is running?

if you can think of any other Veeam backup deployment im also open to that, if it provides tangible advantages.

Best Regards,
M.Rasoolinia

[HannesK]

Hello,
and welcome to the forum.

Foreword: the recommended reference architecture has dumb disk storage as primary backup target for restore performance reasons. Inline deduplication appliances (like StoreOnce) should only be used as backup copy target. In general, deduplication appliances are rarely ever worth the money vs. XFS / REFS space savings, but that's a different story

1. It depends on the number of machines / amount of data you want to back up. As rule of thumb, we usually say "up to 1000 machines per physical proxy". In the end, you have to do the math with your environment values.

2. No, that's direct.

if you can think of any other Veeam backup deployment im also open to that, if it provides tangible advantages.

see my foreword... Replace StoreOnce by Apollo servers with XFS / REFS and you will have faster backup & restore while saving a lot of money StoreOnce to tape will be slow...

Best regards,
Hannes

[ledvanb]

Hi Hannes,

i have to pinpoint the fact that my company has already bought the appliances , im just trying to deploy them as smooth as possible and get the best throughput i can out of them. one thing that is raising my concern is that im installing all the veeam components in an all in one arch on a single physical server.

1.i want to know when im replicating data between storeonce appliances with backup copy /catalyst copy job, will the datamovers on proxy servers be involved between two sites or they will not be involved since everything is orchestrated through catalyst copy?(i know when we add a dedup appliance such as Storageonce an agent will be installed on gateway server)

[HannesK]

in an all in one arch on a single physical server.

that will be fine for many scenarios. how many machines do you have, how much data? was the design done by one of the Veeam engineers, a partner or yourself?

everything is orchestrated through catalyst copy?

yes, that's what I tried to say in 2. with "direct"

[ledvanb]

Hi HannnesK,

thanks again for your prompt reply. we will be using3*Dl380 G10 servers(2*24 cores CPU+256GB RAM) physical server , one for each Site. Each sit will b having its own Backup server, which all be managed by enterprise backup manager. we have around 500 vm on each Sites ,the Store-once is the secondary backup target , reading data from storage snapshots and will replicate them to another Store-once in another site. Im wondering if Veeam still does not provide FT(zero downtime) capability for Backup server in physical environments?

[HannesK]

Hello,
48 cores for 500 VMs sounds good. Without knowing detailed requirements, I would be optimistic that this will work fine

No, there is no Veeam cluster or anything like that. Several approaches are described in this thread

Best regards,
Hannes