Veeam 11 Implementation / Configuration - New Customer Questions

[paps40]

Veeam,
I am working on a project to replace EMC Avamar / Data Domain with Veeam / Exagrid. I have a new vmware 6.7 SSD vsan cluster that I built prior to starting this project. My plan was to deploy veeam as a vm and 1 proxy per each esxi host. I have been looking over the veeam documentation but wanted to ask the forum about some best practice configuration ideas. Our main Veeam VM will run Veeam B&R and Veeam One using the SQL Express DB. Environment has about 300 VM's so not that large. Now is the time to build and configure this using best practices before we migrate off the old backup software and put this in production.

1 How does the Veeam software utilize the CPU's?
For example, would 1 socket – 8 Cores or 2 sockets – 4 cores each be better? What gives the best performance in the veeam software?

2 What are some best practice designs for a new implementation?
Example, 2GB of RAM per each Job.

3 How should the proxies be built?
Similar to main veeam server or lower specs.

4 Veeam Permissions / Group Policy
I read that Veeam needs local admin rights on all servers for application aware processing and restores. I have veeam service accounts setup for that.

What else needs to be in the group policy?
All I can find in the documentation is this, Logon as a batch job granted and Deny logon as a batch job not set
https://helpcenter.veeam.com/docs/backu ... =110#rptcb

Is this all that I need? Anything else?

5 Networking
I am not going to use Jumbo Frames as I can not easily configure that end to end.
Networking in place is 10GB. Is this going to make a big difference 1500 MTU vs 9000 MTU on a 10 GB Network?

6. Other?
Anything else that I am missing or should add?

Thank you in advance for any and all help!

[ronnmartin61]

I'd recommend you review our current best practices guide - https://bp.veeam.com/vbr

[PetrM]

Hello,

I guess that you can find answers to the questions 1 and 2 on the page 8 of our release notes. The main point is to correlate available resources to the number of concurrent tasks you're going to process simultaneously on the same proxy.

Speaking about the question 3, I'd suggest to keep in mind that Direct Storage Access is the most performant transport mode. If it's not available, you may consider HotAdd mode and virtual proxies. Also, take a look at storage integration as it allows proxy to read data directly from the storage snapshot and to shorten VMware snapshot lifetime, thus to reduce impact of snapshot commit.

It's difficult to comment on the question 4, just follow the instructions from our technical documentation.

Jumbo Frames could be useful but I don't think that you should change defaults, what's the point to optimize it if the current configuration does not create a critical "bottleneck" and does not impact your backup window? Anyway, you may refer to this topic as well.

Other: The best practices guide suggested by Ronn is definitely one of the best sources of information. Also, general security considerations may be consulted on this page.

Thanks!