Skip to main content

Advanced Server Disappointments

Thread needs solution

I've been using previous versions of Acronis to backup up workstations and servers. It worked well allowing fast incremental backups and easy recovery into VMs etc.

However, there was one ongoing issue. The archives grew to be very large and required manual consolidating at regular intervals. I bought Backup and Recovery 10 Advanced Server/Workstation hoping that the Storage Node and Deduplication would allow me to automate these tasks on the server.

Here's what I found:

1. Deduplication is too slow to be usable

I've read the KBs and tried SSD/regular SATA disks on high specserver, but the indexing can take days. Have a look through the forums - there are lots of other users who have come to the same conclusion. It's not usuable in even a very small setup - a single server and a few workstations each with around 100GB data.

Technically, the problem seems to be the Vault DB. It must be attempting to a large number of flushed write operations. On an SSD the drive was 95% utilized but only managing about 1MB/s. Moving to a 1TB 7200RPM SATA drive improved the throughput to 5MB/s and reduced the drive utilization.

2. Deduplication indexing locks the storage node until complete

The slow deduplication speed wouldn't really matter if it didn't block subsequent backups. Couldn't it just be a background garbage collection type task? Why not allow backups to proceed in the normal manner on the storage node and then cleanup them up as required when the node was idle. The indexing job could start/stop as required.

3. Can't run standalone cleanup tasks on the storage node

After coming to the conclusion that the deduplication feature I purchased was unusable, I accepted that the archives were going to be larger than they needed to be. I then set about creating server-side tasks on the storage node to do regular consolidation of the backups to at least limit the total size of the archives over time.

There doesn't seem to be a way to do this. You can either cleanup before/after a backup or on a schedule but only in a task on the originating machine. I assume consolidating could take some time and it's going to require the originating machine to be active on the network while the cleanup task executes. We've got laptops that are not always on the LAN and the cleanup task can't run while they are away.

Hopefully, I've misunderstood the product and there's a way around these issues. It doesn't seem like my requirements are very special. We just want an easy, automated way to limit the size of the backup archives with slowing down the normal day to day backups.

0 Users found this helpful

I don't have experience yet with cleanup tasks, but I agree with your analysis of deduplication and indexing. It is a neat feature, but far too slow, even on fast hardware. Everyone complains about the indexing task, but has anyone tried an export or a restore? They are even worse!

I ran a simple test on my storage node (Dell PowerEdge server dual quad-core, 8GB memory, RAID 5 array of 7200 RPM SATA drives on a PERC 5i card).
I created two centralized managed vaults. One deplicating and one non-deduplicating. The Vaults are on one RAID array, the databases on another.

I did a full backup of a 40GB drive to each vault. Yeah, the deduplicated vault went off and indexed for, it seemed, forever. But when that was done, I tried an export (to test making a backup to take offsite). The non-deduplicated vault created the exported backup in 18 minutes. The export from the deduplicated vault is still running, but reports it will take 21 hours!

I contacted Acronis Support. The agent went over the machine and found nothing odd, and eventually concluded that the level of performance is all I can expect with deduplication turned on, but assured me they are fixing it in the next release :) Here's the log where the agent recommends that I don't use deduplication.

> Matiuj Zaman: I have gone ahead and discussed this issue with my expert team and got the information that during the export task the reverse process of de-dupe happens and since it will no longer be in a de-dupe vault and it needs to have all the files on it including the duplicate files, so it will take longer than the backup process. Our development team is working on it and we expect the fix with next Acronis Backup and Recovery
> Ed Kuzemchak: So, this level of performance is expected with Export?
> Matiuj Zaman: For now I would request you to use Non-de-dupe Vault to perform this task.

So, I'm bailing on deduplication, because disks are not that expensive.

Hello all,

Thank you very much for your feedback. I would like to offer you additional assistance with deduplication.

We have a best practices Knowledge Base article available here that can help you to setup deduplication in the most efficient way.

I have forwarded your comments to our program management team and you can always submit your feedback from this link.

If you have additional questions or any other issues, please let me know.

Thank you.