Question about Tape backup size and backup performance.

[Mike102]

Hi,

I have a "forward incremental forever" backup job which is configured to keep 3 restore points. There is also a Weekly Tape job to archive the backup to tape. In the Repository folder for the job, there is a 3.3TB vbk file and two vib files, 150GB and 106GB. The Tape job has the option “Archive Incremental backups to Tape” ticked. The Tape job processes 6TB of data, and transfers 6TB of data. Why does the Tape backup take 6TB? I appreciate it is creating the synthetic full during the backup. Also we are using an LTO6 library over 8Gb fibre-channel switches, 8Gb hbas, but the throughput average is 56MB/sec, so takes 32 hours. The job says the target (Library) is the bottleneck, but it shouldn’t be. We use Veeam 9 update1

Thanks

Mike

[veremin]

Can you check what accounts for this amount of data? Say, additional backup files you haven't thought about or something else? Moreover, isn't repository a Windows Server 2012 (r2) with Data Deduplication role enabled by any chance? Thanks.

[Mike102]

Hi Vladimir,

OS dedup is not enabled in 2012. I rarely use that feature, as it kills the performance when archiving to Tape. I checked the backup job again, and noted that I had set Compression to extreme. Looking at the report history, sometimes the compression ratio is 2.8x. So I think that explains the 6TB figure, which goes to Tape as well as the poor backup performance. I have another tape job, that uses the same drive, which archives multiple jobs, with compression set to optimal (default). Also those jobs are not forever incremental, so there is no synthetic full process during the archive to tape. For this job, the throughput is 120MB/sec

Mike

[veremin]

I'm still wondering how 3 TB of data can occupy about 6 TB on tapes. Did a tape job not only archive original .vbk plus .vib's, but also synthesize a virtual full backup? Thanks.