Backup performance on NX-1050

  • 11 July 2014
  • 5 replies
  • 1174 views

Badge +4
I currently have 2 x NX-1450, both are running NOS 4.0.1 and I'm using Hyper-V. Everything is linked up to 10Gb and I'm using a physical veeam server outside of Nutanix to back up both clusters, he is linked up to 4Gb. We are currently suffering from a bad backup performance, let's say something around 30-40Mb/s processing rate for one VM that is on a node that isn't doing anything at all.

If I perform the same tests on an NX-3360, I can easily get up to 160-180MB/s processing rate, so that kinda rules out my network/configuration.

Is there anyone out there running NX-1050 with vmware or hyper-v and Veeam and that can share their configuration/processing rates?

This topic has been closed for comments

5 replies

Userlevel 4
Badge +19
Hi

What is your power policy set too?
Badge +4
Ever since NOS 3.5.4 came out, the power plan is set to high performance automatically, I checked it myself for all my nodes
Badge +5
Having worked with Veeam B&R across many types of hardware platforms (Inc lots of Nutanix), I'd typically say that 30-60Mbps is a normal processing rate for backups.
There are too many variables to list here, however to list a couple:
  • Backup Type: Inc / Rev Inc / Full
  • Backup Mode: Hot-add / NDB
  • Compression / Dedupe Levels
  • Backup Target
If you're running a backup of a VM where the majority of change data has been migrated to the SATA drives, expect a slower rate than an incremental backup of a VM where the incremental data is still stored in flash.

My questions back would be:
  • What did Veeam report as the bottleneck?
  • Are you having issues exceeding your backup windows?
Badge +4
Hi and thank you for your reply, we are currently a lot further in our setup and I wanted to share some of our findings.

We are using the NX-1050's together with Veeam B&R V8, i can see a processing rate of 150-170MB's for incremental backup jobs's, currently 16 vm's on our cluster. Also to shed some light on the SSD - SATA usage on the storage pool, currently running 32% in SSD and 68% in SATA.

Backup job settings:
on-host backup - repository is going to an FC LUN connected on the physical Veeam Server, we are using incremental backups with synthetic fulls. We have enabled inline data deduplication and compression has been set to optimal. Guest quiescense, application-aware and indexing is currently disabled.

Veeam is always reporting the bottleneck as being the source, I hope i will not be exceeding my backup window since i am currently doing 1:45 to 2 hours for this job.

Anybody else willing to share their experiences?
Userlevel 6
Badge +29
I haven't run into that exact configuration, but we can certainly dig deeper.

I'd hate to guess and have you run in circles, I'd recommend opening a ticket with Nutanix support. They can help review your configuration end to end and, if needed, do a real time trace during your backup window (called perf collect), to see if there is anything redlining within the Nutanix storage stack.