-
- Influencer
- Posts: 14
- Liked: never
- Joined: Feb 03, 2018 10:11 pm
- Full Name: I Hussain
- Contact:
Storage Optimization setting for Replica jobs
Hi
If the replication jobs are run from a virtual VBR server (that is hosted on the same ESXi server for replicas) at the DR site then what should the storage optimisation setting be set to?
If the replication jobs are run from a virtual VBR server (that is hosted on the same ESXi server for replicas) at the DR site then what should the storage optimisation setting be set to?
-
- Veeam Software
- Posts: 1922
- Liked: 636 times
- Joined: Sep 25, 2019 10:32 am
- Full Name: Oleg Feoktistov
- Contact:
Re: Storage Optimization setting for Replica jobs
Hi,
It would rather depend on replication strategy you're using (on- or off-site) and network bandwidth/throughput between sites than VBR server placement. VBR server hosted at the target replication site is helpful for triggering failovers if the source site is down, whereas data transfer traffic still goes between source - target entities.
As far as I understood, you replicate off-site, right?
How fast is the network link between sites?
Do you use WAN accelerators or transfer data through proxies?
Thanks,
Oleg
It would rather depend on replication strategy you're using (on- or off-site) and network bandwidth/throughput between sites than VBR server placement. VBR server hosted at the target replication site is helpful for triggering failovers if the source site is down, whereas data transfer traffic still goes between source - target entities.
As far as I understood, you replicate off-site, right?
How fast is the network link between sites?
Do you use WAN accelerators or transfer data through proxies?
Thanks,
Oleg
-
- Influencer
- Posts: 14
- Liked: never
- Joined: Feb 03, 2018 10:11 pm
- Full Name: I Hussain
- Contact:
Re: Storage Optimization setting for Replica jobs
Hi Oleg
Yes that's right. Replication is to an offsite ESXi host over a VPN connection to our DR site.
Here's a brief summary of the setup.
- The physical VBR server in the Production site utilises direct-SAN backups. Backup copies are sent to a NAS in the DR site. This VBR server is also the VEM and is purely for backup and backup-copy jobs only. These are run once a day.
- The virtual VBR server in the DR site is solely for running replication jobs and for initiating failovers when the production VMs/site are down. These jobs are run twice a day.
- Replica seeding is enabled on all jobs.
- There is a 100Mbps (up/down) link between the 2 sites. The bandwidth reports show that this never peaks above 50-60Mbps even with multiple backup-copy or replication jobs run simultaneously.
- Both VBR servers are the backup proxies. The WAN accelerators are not being used as I noticed the backup-copy and replication jobs take longer to complete than with the direct transfers.
Hope this helps.
Yes that's right. Replication is to an offsite ESXi host over a VPN connection to our DR site.
Here's a brief summary of the setup.
- The physical VBR server in the Production site utilises direct-SAN backups. Backup copies are sent to a NAS in the DR site. This VBR server is also the VEM and is purely for backup and backup-copy jobs only. These are run once a day.
- The virtual VBR server in the DR site is solely for running replication jobs and for initiating failovers when the production VMs/site are down. These jobs are run twice a day.
- Replica seeding is enabled on all jobs.
- There is a 100Mbps (up/down) link between the 2 sites. The bandwidth reports show that this never peaks above 50-60Mbps even with multiple backup-copy or replication jobs run simultaneously.
- Both VBR servers are the backup proxies. The WAN accelerators are not being used as I noticed the backup-copy and replication jobs take longer to complete than with the direct transfers.
Hope this helps.
-
- Influencer
- Posts: 14
- Liked: never
- Joined: Feb 03, 2018 10:11 pm
- Full Name: I Hussain
- Contact:
Re: Storage Optimization setting for Replica jobs
Also the virtual VBR server in the DR site is connected to the NAS (which holds the backup copies) over a 2Gbe (1Gb x 2 etherchannel) connection on a separate vSwitch to the ESXi management and VM networks.
-
- Veeam Software
- Posts: 1922
- Liked: 636 times
- Joined: Sep 25, 2019 10:32 am
- Full Name: Oleg Feoktistov
- Contact:
Re: Storage Optimization setting for Replica jobs
In your case, If you have quite powerful proxies, I'd advise to use "WAN target" option among others as then source VM image is to be split into blocks 256 kb each during job processing meaning smaller increments produced. So, the smaller the block size of each changed data block found the lesser amount of data is to be transferred to the DR site.
Keep in mind, though, that it might lead to an additional load on proxy CPU and reduced job performance.
Also, make sure you enabled CBT and check this page for more info on data traffic optimization.
Best regards,
Oleg
Keep in mind, though, that it might lead to an additional load on proxy CPU and reduced job performance.
Also, make sure you enabled CBT and check this page for more info on data traffic optimization.
Best regards,
Oleg
-
- Veeam Software
- Posts: 21074
- Liked: 2115 times
- Joined: Jul 11, 2011 10:22 am
- Full Name: Alexander Fogelson
- Contact:
Re: Storage Optimization setting for Replica jobs
Also, you could avoid sending the data three times over the WAN link by performing more frequent backup copies and using replica from backups directly in the remote site. In this case, however, it is better to have a single Veeam B&R instance, otherwise, some sort of creative workaround would be required.
-
- Influencer
- Posts: 14
- Liked: never
- Joined: Feb 03, 2018 10:11 pm
- Full Name: I Hussain
- Contact:
Re: Storage Optimization setting for Replica jobs
Thank you guys for the clarification and details.
Just a related query if I keep the storage optimisation setting as "WAN target" on the replication jobs then how can I tell how much bandwidth is being consumed during the job run - does VBR report this? Also how would I know if it's beneficial to enable the WAN accelerators or not?
Just a related query if I keep the storage optimisation setting as "WAN target" on the replication jobs then how can I tell how much bandwidth is being consumed during the job run - does VBR report this? Also how would I know if it's beneficial to enable the WAN accelerators or not?
-
- Veeam Software
- Posts: 21074
- Liked: 2115 times
- Joined: Jul 11, 2011 10:22 am
- Full Name: Alexander Fogelson
- Contact:
Re: Storage Optimization setting for Replica jobs
Basically, you just test and compare the amount of data transferred in both cases. This information is available in the job session log.
-
- Influencer
- Posts: 14
- Liked: never
- Joined: Feb 03, 2018 10:11 pm
- Full Name: I Hussain
- Contact:
Re: Storage Optimization setting for Replica jobs
I see...but isn’t it possible for VBR to show in real time the wan bandwidth consumed (e.g. in mbps) whilst a job is running or after it has completed?
-
- Veeam Software
- Posts: 21074
- Liked: 2115 times
- Joined: Jul 11, 2011 10:22 am
- Full Name: Alexander Fogelson
- Contact:
Re: Storage Optimization setting for Replica jobs
Yes, it shows the stats while processing - that's why you need to run some tests and monitor those.
-
- Influencer
- Posts: 14
- Liked: never
- Joined: Feb 03, 2018 10:11 pm
- Full Name: I Hussain
- Contact:
Re: Storage Optimization setting for Replica jobs
Thanks
It seems that the jobs do take longer to run when using the built-in WAN accelerators instead of direct data transfer.
I have noticed that VBR is only utilising around 50% of the available bandwidth (50Mbps out 100Mbps) when the backup copy and replication jobs are running. The bottleneck for the jobs is network. Could this be due to the storage optimisation setting set to WAN? Isn't there any way to force VBR to use all/close to the available bandwidth?
It seems that the jobs do take longer to run when using the built-in WAN accelerators instead of direct data transfer.
I have noticed that VBR is only utilising around 50% of the available bandwidth (50Mbps out 100Mbps) when the backup copy and replication jobs are running. The bottleneck for the jobs is network. Could this be due to the storage optimisation setting set to WAN? Isn't there any way to force VBR to use all/close to the available bandwidth?
-
- Veeam Software
- Posts: 3276
- Liked: 529 times
- Joined: Aug 28, 2013 8:23 am
- Full Name: Petr Makarov
- Location: Prague, Czech Republic
- Contact:
Re: Storage Optimization setting for Replica jobs
Hello,
In case of direct transfer without WAN-accelerators, you may try to increase "Use multiple upload streams per job" value to better utilize high-latency link.
For WAN-accelerators, you can control number of streams in properties, may be it's worth testing in your case.
Thanks!
I don't think so... The "bottleneck" Network means that job activity time is mostly spent on data transfer through the link between sites and this parameter does not affect speed of data transfer.ihussain wrote:Could this be due to the storage optimisation setting set to WAN?
In case of direct transfer without WAN-accelerators, you may try to increase "Use multiple upload streams per job" value to better utilize high-latency link.
For WAN-accelerators, you can control number of streams in properties, may be it's worth testing in your case.
Thanks!
Who is online
Users browsing this forum: Google [Bot], neilp and 23 guests