Veeam immutable repository for SMB's

[DVelez]

I’m currently evaluating the Veeam hardened repository included in the latest VIA release, with the goal of recommending it to my SMB customers (most have 15 or fewer VMs). During testing, I’ve run into a few questions:

1. Does the server require a RAID controller, even for small environments? If that's the case, should it include a cache module?
2. If later I add an additional hard drive later, what’s the process for expanding the volume? Can the VSA be managed similarly to other Linux distributions?
3. I’m considering the HPE MicroServer and Lenovo ThinkSystem ST50 for these deployments. Are there other hardware options you’d recommend for very small environments?

Thanks in advance!
DV

[pybfr]

Hello Daniel
1. Yes and Yes
2. You can't currently
3. Not familiar with this type of configuration, sorry.

[DVelez]

Hello pybfr!
Is the cache module meant only for better performance, or does it offer other benefits? In this case, my customers transfer a few gigabytes per day, so they are not looking for performance gains with the repository.

[pybfr]

The cache module helps performance but its main role is to protect data integrity in case of crash or power failure.

[DVelez]

Would it help to configure the controller to use write-through instead of write-back? As mentioned earlier, this is intended for small customers with light requirements.

[pybfr]

Write through is going to be 10 to 15x slower than write back. But If your controller does not have a battey to protect the cache (BBU) you shouldn't use WB at all....

[DVelez]

Given the small amount of data these customers back up daily, performance degradations will have no noticeable impact on them. Ultimately, the key question is: Can I use a certified server with a cache-less controller configured for write-through mode?

[pybfr]

The question is not really about performance, it is more about the importance you give to your backups.

We do not certify servers so you are free to do what you want, but be aware that is something happen during your backup windows there is a good chance you end up with data corruption.

Personally, this is something I would avoid, but everyone has different priorities

[Gostev]

Power loss during write may leave your backups corrupted due to the RAID "write hole" phenomenon. So you just need to decide if potentially unrecoverable backups is an acceptable risk that is worth the added price of a RAID controller with battery/capacitor backed write cache.

[DVelez]

I’m not a storage expert, so this might be why it’s the first time I’ve come across the term “write hole.” I’ll do some research on it. Thank you both for taking the time to answer my questions.

[pybfr]

With the increase of disk size, disk rebuild in a RAID may take several days, while adding pressure on remaining disks, there is a risk of the whole array goes down with complete data loss.
That's why RAID 5 is no longer recommended for critical application, RAID 6 being the standard configuration, and RAID 60 when the amount of disk in the array increases...

[karsten123]

+ at least one HS

[pybfr]

Excellent point!