Hyper-V backups - Agent based Accelerated Vs Snapshot based backups

Seems no one wanted to comment on this so here are my thoughts if they help.... I know you have used the term Accelerated but my line of thinking covers de-dupe in general

Hyper-V snapshots the whole hyper-v server when you back up a client so is perhaps not as efficient as a client based backup

Hyper-v does not integrate with applications in the same way as VMware so you cannot do Exchange / SQL / SharePoint backups via hyper-v

Any application backups need to be done in the traditional manner anyway for a pure agent backup

Agents cannot be done using accelerator but they can use client side de-dupe which minimises the network bandwidth used

Accelerator backups of large file systems are very efficient and fast so great to use, especially combined with client side de-dupe and allow you to use ALL_LOCAL_DRIVES so you get the Shadow Copy Components backed up - which you do not get with a Hyper-V backup - but for any client that is just a general small to medium sized server then the hyper-v method is probably favorite

Summary:

Small / medium servers work well using Hyper-V

Large Servers are possibly best using accelerator - especially as these are usually the ones that need the most restores so it is handy that they already have an agent instaled

Apps Servers than need an agent are best using client side de-dupe

If you may need to recover the Shadow Copy Components then go for Accelerator

Just a few thoughts.....

There are as many as 100 VMs hosted on 7 CSVc and 4 nodes. The servers are powerful blade servers with 128GB RAM and 16 cores each. Doing accelerator backup will increase the load on the appliance as it also needs to perform synthetic backup process after collecting deltas to create a Full backup image. I have seen the appliance hung sevral times and Symantec SE is saying that it can happen because of accelerator being used heavily. We have two 5230 and the Hyperv Hosts are connecting using 10Gb ethernet.

In this scenario, do you think snapshot method will be benificial than accelerator. Also if we enable source dedup, will it help as much as accelerator to reduce the time taken to complete the backups.

Thank you.

Have you incrased the worker threads on your appliances to cope with using accelerator?

By default it uses 64 which is not enough - increase it to 128 or 256:

/disk/etc/puredisk/contentrouter.cfg

The "WorkerThreads" value

Needs a re-start to take effect

Hi Mark,

Thanks again for your inputs.

Same was discussed with Symantec SE who is responsible for this perticular account. He believes the vaule should not be changed unless suggested by support. And you know how long support takes to collect and verify logs. :(

As of now we do not have any updates from support.

The upgrade to 2.6 is due and I was wondering if this vaule has been increased in 2.6. Any idea?

Thanks Mark. I shall let you know how it goes after we upgrade to 2.6