I have raised a call for this Case #05487927 but would appreciate any help.
Setup
Veeam v11
4 node Dell EMC Isilon cluster
Each node has 2 x 10GB network connectivity
4 Backup Proxy servers
Isilon cluster added in Storage Infrastructure and NAS filer
When adding in the Isilon cluster to Storage Infrastructure I selected the 4 Backup Proxy servers
A backup job has been tested to backup an Access Zone on the Isilon cluster
The shares in the Access Zone are SMB
The access zone is 65TB in size
The job started but after 15-20 minutes a node on the Isilon cluster maxed out the CPU, this continued and the job was canceled
Re-testing the job produced the same issue
I would like to ask for help with the configuration of the setup to enable the streams to be spread across the nodes in the cluster. I did think that having 4 Backup Proxy servers allocated would achieve this.
If the resource consumption at the source side is a problem you can change the File Share Processing Settings and limit how aggressively file proxies can read from your Isilon servers via Backup I/O control. Try to lower the value and check the consumption on the Isilon's end. Thanks!
I have reduced the backup I/O control so as not to cause performance issues with the Isilon.
The real issue is that the job is only processed through one of the Isilon nodes. Node 2 is showing high CPU usage and disk I/O, whereas nodes 1,3,4 are showing minimal activity. I would like the processing on the Isilon to be shared across the nodes in the cluster.
I allocated four backup proxy servers to the Isilon cluster when I added it to the Veeam Infrastructure. I can see that when I run the file backup job that VeeamAgent processes do appear on all four backup proxy servers.
When I added the Isilon cluster into the Veeam Infrastructure I did put in an FQDN.
Veeam Support has asked for logs but when I try to export them I am getting an error message.
and saw the following information posted by Andreas Neufert:
"Isilon usually uses DNS round robin to spread the load accross multiple nodes (or external DNS load balancers). In both cases out Repository Gateway Server will get an IP address from one of the controllers, will cache it and use only this controller for the workload. Therefore it is best to use multiple Repository Gateway Server (1:1 with the number of Isilon nodes maybe) and use individual folders for each of them within the share (do not use a central folder or just the share for all of them). That way Veeam will not only use one of the nodes for IO"
This certainly seems to be what is required but I'm unsure why it isn't load balancing.
Unfortunately nas job does not know anything about Isilon round robin load balancing. Can only add you vote to this feature request but meanwhile the best workaround would be to backup from each node separately.
After a lot of testing, I did discover that the CPU % rises when the share files and folders are being calculated. Once that is complete the CPU % does drop down again. Is there any way to skip this calculating process?
A job with just one share added stabilised at CPU % of 5% once the calculation was complete. I have started another job with three shares added and the CPU % is 15% once the calculation was complete.
I would like to run a job that backs up an Isilon Access Zone with approx 50 shares. My concern is that the CPU % climbs in line with the number of shares in the job.
All of the testing has been conducted with native Isilon integration to Veeam.
Adding in additional shares to a file backup job for Isilon does indeed increase the CPU % on the Isilon. I have a six share Isilon backup and one of the shares is still calculating the files and folder after an hour and 15 minutes. If I reduce the Backup I/O control then the backup becomes too slow. If I increase it then the CPU % rises. It does not appear to be a useable solution for backing up Isilon shares. The aim is to select an Access Zone and all of the shares in that AZ will be backed up, but when doing that I still had shares calculating the files and folders after 20 hours and had to cancel the job due to an unresponsive node. As all of the calculation is done on one Isilon node then the CPU goes to 100% and affects user access.
I have the same problem with all backup jobs coming to the same Isilon node.
"Therefore it is best to use multiple Repository Gateway Server (1:1 with the number of Isilon nodes maybe)" - from post above.
I see, does it mean, that I have to build different repository servers and split repositories between different shares? What is repository Gateway Server? Is it just a Repository Server?
What about proxy servers, are they connecting to Isilon nodes, or request is coming from repository server and then DNS Round Robin works?
