Hi, I am sorry but I don't understand what are you saying here. Can you please explain your question or what are you trying to achieve in different words?
skydok wrote:But suppose i have 1001 Cores
This is not realistic. I am yet to see the server with even 100 cores used as a backup proxy in the wild, but 1001 cores configuration? Perhaps we will see those in 5-10 years from now only. Regardless, this will be huge overkill for a backup proxy (your bottleneck will be storage and fabric speed).
Gostev, i know about the 1001 Cores, but it was for the example purpose.
I want to know if there is a HardCoded Limit for Parallel VMs Backups in the Same job / Across Jobs.
i want to know if its possible to backup 1001 VMs in the same time writing to the repository.(lets assume its an all Flash array with Infiniband)
I am trying to confute a solution for a customer who is currently saying that he needs more than 100 VMs in the same time to be processed.He Got EMC Networker and says that backup trough VADP has limitation from VMware which is only 100 .
There is no hard code limit for the number of tasks on the repository if you uncheck this option in the GUI. To process more than 100 VMs at the same time you would need multiple proxy servers and good storage disks on the repository server. Thanks!
Vitally ,Thanks for the assurance . I just wanted to be sure that with Veeam i don't have any hidden limitation like in other Applications from competitors.
dellock6 wrote:There is still an implicit limit of 1 parallel job per core, regardless it's virtual or physical, isn't it?
This is only a recommendation. We will warn you if you attempt to set the number of max tasks to more than your number of cores, however this is about it. Meaning, you can still go ahead and set max tasks on 1 core proxy to 1000 for example, if you'd like
I got the following information that the limit is 100 because of the below Marked Parameter ,
and some other Vendor says that he is limited to that 100 because VADP is limited.
IMHO: When you need to backup such a number of VMs at the same time (or within seconds) then the best is to rely on hardware snapshots. It allows you to freeze many VM at the same time and then process them later.
By the way, I see two reasons why your customer might need to process those VMs simultaneously:
- 1 : the backup window is constraint and you need to handle as much VM as you can to fit in.
- 2 : many VMs are working together for the same application and he needs them to be frozen at the same time to maintain data consistency in case of restore
Perhaps I don't understand the point here...
Veeamizing your IT since 2009/ Veeam Vanguard 2015 - 2023
And apart all the good points coming out in this discussion, I think a scale-up design for a proxy is not the best choice, so the per-proxy limit is really not a limit. When the infrastructure is at this size, the only viable approach is scale-out, and that's mainly for two reasons:
- it easier to distribute the load of the jobs among several proxies rather than using a huge single proxy: even if we assume the total cpu and ram is the same, there are other problems as for example the storage connection bandwidth on a single connection of a huge single proxy compared to multiple proxies using different links
- failover and high availability: with several proxies jobs can be completed even if a single proxies crashes or have poblems
My 2 cents.
Luca Dell'Oca Principal EMEA Cloud Architect @ Veeam Software
