VMworld 2012: SPO3292 – Veeam Backup & Replication: Tips and Tricks

VMworld 2012: SPO3292 – Veeam Backup & Replication: Tips and Tricks

Session abstract

While Veeam works great right out of the box, it also provides a lot of options for configuring and tuning your backup infrastructure. Anton Gostev (@Gostev) and Doug Hazelman (@VMdoug) share expert insights and practical advice for VM backup. What’s the best and safest way to configure direct SAN backups? When is network processing mode a better choice than hot add? How to efficiently write Veeam backups to deduplicating storage, to another site, or to the cloud? What are the best practices for deploying your backup management server? Whether you currently use Veeam Backup & replication or just want a better understanding of image-level backups of virtual machines, this session on advanced disaster recovery and business continuity is for you.

The session

This session started of with Doug Hazelman making an introduction about how Veeam 6 introduced backup proxies and how it works. He then gave the word to Anton Gostev who went through tips and tricks in an interesting way: “The good, the bad, the ugly“.

Scaling your backups
This was the first point in the session and it answered some frequently asked questions but also gave some basic tips when you start working with Veeam (or if you already are which you should do).

I am not going to list them all but some important ones:

  • Disable the default proxy on the management server
  • Allocate enough RAM for job manager processes
  • Don’t go nuts on backup proxy servers, this will create too much load on both storage and network
  • Be carefull with the reversed incremental backup model
  • Limit your concurrent jobs to a reasonable amount, don’t kill your storage or backup repo

Backup repositories
These are the most commonly reported bottleneck. They either be a Windows or Linux server (they can even be the same as (one of) the backup proxy servers). If you can afford it: use RAID10!

The session went one with good, bad and ugly things about each backup model:

  • Direct SAN Access
  • Hot Add
  • Network mode

Direct SAN Access is probably the best because of several reasons:

  • Fastest processing mode
  • Least impact on your production!
  • It doesn’t impact your consolidation ratio

A disadvantage about this that it only support block storage (iSCSI or FC) and if you have a FC SAN it is required that you use a physical backup proxy server.

Some tips/tricks:

  • Present VMFS LUNs as read-only
  • Disable automount on your backup proxy servers!
  • Disable disk management on the proxy servers!
  • Disabling MPIO might increase performance.

Another great tip if you are using iSCSI SAN is that you can tweak TCP/IP on your backup proxy. According to Anton this can improve performance by 2 times the normal handling. You can change this using the following command:

netsh interface tcp set global autotuninglevel = disable

Hot Add is a different story compared to Direct SAN Access. It’s easier to setup and it supports all types of storage (NFS, iSCSI, FC, local storage). A great tip is that you can use any Windows for this so re-use you current Windows servers and save on the licensecost ;-). The problem with Hot Add is that it slower because the process itself takes a while.

The ugly part about Hot Add is that you might run into snapshot problems due the locking mechanism. Another thing is that you must disable CBT on your backup proxy (very important).

Some tips/tricks:

  • Add extra virtual SCSI controllers to your backup proxy server
  • Keep vSphere and Veeam up to date
  • Avoid cloning a backup proxy VM

Finally the session had a small talk about Network Mode. This is by far the slowest method you can use if you are on a 1GbE network. Keep in mind that a restore of a full backup can take a very long time. A big tip is that if you have to use Network Mode: upgrade to Veeam B&R 6.1 as there is improved network location awareness with this release (and higher).

A final (big) tip: disable VDDK logging on your virtual machines! This can easily save 5 minutes (or even more) of processing time.

Anton also talked about deduplicating storage. He mentioned that the dedupe with Windows Server 2012 is amazing and it’s one of the best they ever saw (and it’s free).

If you currently have storage with inline deduplication keep in mind that this makes vPower slow. Disable it if you can and find another solution. Another tip is that you can reduze the block size (WAN: 256KB / LAN: 512KB) but keep in mind that this might impact your backup performance.

The future

At the end of the session Doug talked about the new upcoming 6.5 release which includes several great tools.

  • Veeam Explorer for Exchange
  • Veeam Explorer for SAN Snapshots

More information about 6.5 can be found on their blogpost.

This session was one of the best I attended, this was really useful for people who start with Veeam but also for people who have been using it for a while now.

Niels Engelen
Working as a Global Technologist in Product Strategy for Veeam Software with an interest for anything virtual and cloud. Niels is as a VMware Certified Professional (VCP-DCV 4, 5, 6) and a Veeam Certified Architect (VMCA). Niels also gained the VMware vExpert award (2012-2017) and is an ex-PernixPro.

11 thoughts on “VMworld 2012: SPO3292 – Veeam Backup & Replication: Tips and Tricks

  1. Hi Niels, nice article, thanks for it!
    Only correct the autotuning command, “autotuninglevel” has no space in it. Just in case someone wants to copy/paste the command.

Comments are closed.

Comments are closed.