Direct Restore to Microsoft Azure | VeeamPN software-defined networking
Post Reply
lukejf
Service Provider
Posts: 54
Liked: 2 times
Joined: Jul 10, 2012 8:15 am
Full Name: Luke
Contact:

Azure Testing

Post by lukejf » Oct 12, 2017 3:17 am

Hi All
We recently build a B & R VM in azure running on a F2 size instance and standard managed disk 4TB. Over our 100MB fibre link using IPSEC we are getting a 15MBps backup copy transfer speed which is pretty reasonable.
We did a test backup which all went great. We have now completed a test restore to Azure from the VM running in Azure. We restored the VM to premium storage but noticed we are getting pretty average restore rates, Only around 18MBps.
I checked the specs and the standard storage (reading from) should tap out at around 60MBps so wondering why the restore performance is so low. Just wondering if any of you regularly run test restores from an azure B & R server and what sort of throughput you are getting.
If you are getting a much higher throughput, what are the specs of your Azure Instance B & R Server.

v.Eremin
Veeam Software
Posts: 15077
Liked: 1133 times
Joined: Oct 26, 2012 3:28 pm
Full Name: Vladimir Eremin
Contact:

Re: Azure Testing

Post by v.Eremin » Oct 12, 2017 1:13 pm

More information regarding the restoration procedure. What type of restore was used? Where backup server and backup repository were located, etc.?

lukejf
Service Provider
Posts: 54
Liked: 2 times
Joined: Jul 10, 2012 8:15 am
Full Name: Luke
Contact:

Re: Azure Testing

Post by lukejf » Oct 12, 2017 10:29 pm

Hi, It was a full VM restore to azure from the backup and replication server stored in Azure in the same tenant. I didn't use an azure proxy as the B & R server is located in the same region so figured it wasn't required.

lukejf
Service Provider
Posts: 54
Liked: 2 times
Joined: Jul 10, 2012 8:15 am
Full Name: Luke
Contact:

Re: Azure Testing

Post by lukejf » Oct 13, 2017 12:10 am

Hi Guys
Just ran another test:
Source Azure B & R Server: F2 instance with Standard Managed Disk
Destination: Same Region, Tenant, Vnet, Premium storage account
Veeam Throughput 32MBps
Source B & Recovery Server Resource utilization
Image

lukejf
Service Provider
Posts: 54
Liked: 2 times
Joined: Jul 10, 2012 8:15 am
Full Name: Luke
Contact:

Re: Azure Testing

Post by lukejf » Oct 16, 2017 2:18 am

Well i have completed some more testing this morning and still not really getting the restore response i was hoping for in azure. So far this has been the throughput of my tests.

Architecture

On Prem
-Veeam Backup & Replication Server

Azure
-Veeam Backup & Replication Server
-Virutal Network Gateway with IPSEC VPN

Test VM Backup
-180 GB Full Backup on Prem Server
-Backup Copy job to Cloud Server

Restoration Procedure
-Import Backup copy job on cloud server
-Direct restore to Azure in same datacentre


Restoration Speed from Azure B & R Server to Same Subscription and datacentre.
From standard Managed Disk to standard account, 36MBps using F2 Instance as B & R Server
From Standard Managed Disk to Premium Account, 36MBps using F2 Instance as B & R Server
From Premium Managed Disk to Premium Account, 37MBps using F2 Instance as B & R Server
From Premium Managed Disk to Premium Account, 39MBps using DS3_v2 Instance Instance as B & R Server

The final test i need to complete is using a veeam proxy server in azure, however i don't really see the point of this unless the restore is being run from the onprem environment.

lukejf
Service Provider
Posts: 54
Liked: 2 times
Joined: Jul 10, 2012 8:15 am
Full Name: Luke
Contact:

Re: Azure Testing

Post by lukejf » Oct 16, 2017 6:20 am

Are any of you guys getting similar speeds or better speeds? If so what is your magic formula as i'm stumped

lukejf
Service Provider
Posts: 54
Liked: 2 times
Joined: Jul 10, 2012 8:15 am
Full Name: Luke
Contact:

Re: Azure Testing

Post by lukejf » Oct 17, 2017 11:18 pm

lukejf wrote:Are any of you guys getting similar speeds or better speeds? If so what is your magic formula as i'm stumped
Is anyone actually using azure as a recovery option? With these speeds i don't see that it is a plausible solution? It would be good to get some feedback from the veeam community so we can decide if we should move on and look at other options

anthonyspiteri79
Veeam Software
Posts: 229
Liked: 31 times
Joined: Jan 14, 2016 6:48 am
Full Name: Anthony Spiteri
Location: Perth, Australia
Contact:

Re: Azure Testing

Post by anthonyspiteri79 » Oct 18, 2017 3:43 pm

Hey Luke.

This is a complex one to respond to with any certainty because millage might vary depending on individual use cases, however the numbers you are putting up are not far off what I have read about and observed in Azure. The Direct Restore to Azure feature is popular and used by a lot of our customers, but you are not going to get super low RTOs compared to say using Cloud Connect Replication from on-premises to a VCSP.

Out of interest what is the specific use case here?
Anthony Spiteri
Global Technologist, Product Strategy | VMware vExpert
Email: anthony.spiteri@veeam.com | Mobile: +61488335699
Twitter: @anthonyspiteri | Skype: anthony_spiteri

lukejf
Service Provider
Posts: 54
Liked: 2 times
Joined: Jul 10, 2012 8:15 am
Full Name: Luke
Contact:

Re: Azure Testing

Post by lukejf » Oct 19, 2017 3:21 am

Hi Anthony
thanks for the reply, We were thinking of using it either as Cloud Connect repo or for customers that are big enough to warrant their own B&R server in Azure.

anthonyspiteri79
Veeam Software
Posts: 229
Liked: 31 times
Joined: Jan 14, 2016 6:48 am
Full Name: Anthony Spiteri
Location: Perth, Australia
Contact:

Re: Azure Testing

Post by anthonyspiteri79 » Oct 19, 2017 3:33 am

No worries Luke.

So those are both valid use cases, but there is a point in where Azure for Cloud Connect stops being viable. It's best suited to smaller data sets or for VCSPs looking to test the waters before investing in their own infrastructure.

If you have access to the VCSP forum...Have a look at this thread for some more background: post257060.html#p257060
Anthony Spiteri
Global Technologist, Product Strategy | VMware vExpert
Email: anthony.spiteri@veeam.com | Mobile: +61488335699
Twitter: @anthonyspiteri | Skype: anthony_spiteri

lukejf
Service Provider
Posts: 54
Liked: 2 times
Joined: Jul 10, 2012 8:15 am
Full Name: Luke
Contact:

Re: Azure Testing

Post by lukejf » Oct 19, 2017 11:32 pm

Thanks Anthony, Some of our workloads we are looking at testing are around the 500GB mark and seem would take a fair while to restore if required, especially if requiring restoration of multiple VM's

jmercado567
Lurker
Posts: 1
Liked: never
Joined: Sep 06, 2018 5:10 pm
Full Name: Jose Mercado
Contact:

Re: Azure Testing

Post by jmercado567 » Sep 06, 2018 5:18 pm

This issue is occurring globally with a lot of systems, protocols etc. It appears there is a global network change where agreed speeds for various protocols are being throttled to desired speeds etc.

For example you can do a speed test on a VM in azure and would have amazing speed results. But try to do FTP, a file copy thru VPN with azure, etc. you will find that speeds are throttled down. I have seen one backup company have avg of 100mbps for file trasfers on their system.

Things are getting throttled and controlled. pretty soon you will be billed at some point for speed wanted per protocol and services on top of your monthly access. guaranteed. when you bring this up with providers and tickets, you totally get what you see here... just finger pointing or totally get ignored. I saw our internet carrier have great speed, then right after the company was purchased, the service average speeds changed. Just less than a year ago downloading Mac OS was a quick download... now its slow, same for MS O365 install packages.

Post Reply

Who is online

Users browsing this forum: No registered users and 1 guest