Tape compression

General Discussion
1

Hello friends,

We are looking at working out our tape compression ratios in our deployment. I attempted to do some calculations using cta-admin, which make sense to me, but the compression ratio I get is very very low. Could someone please have a look and verify that this is indeed the correct way to work out compression? We have 6T tapes

cta-admin --json ta ls -a | jq  -rc '.[] | select(.occupancy|tonumber > 6000000000000) | .occupancy' | awk '{sum+=$1}END{print sum/NR}'

6,009,652,993,055

6,009,652,993,055 / 6,000,000,000,000 = 1.001608832175956

This gives up an average compression ratio of 1:1.001608832175956, which is very low. The specs on the drives suggest a 1:2.5 at best, but 1.1.0016 seems quite low.

We data is restic pack files, which are deduped but not compressed.

Thank you friends :slight_smile:

2

Hi Denis,

my comment would be that deduplication (which removes duplicate blocks) is in fact a particular form of compression / data reduction. So compressing the data further might be difficult.

What ratio do you get if you take the deduplicated data and run gzip on it?

Vladimir

3

Denis,

see this page:

where they say:

Size Reduction Rate

Compression claims to reduce data size to the ratio of 2:1 up to 2.5:1, as claimed by some programs based on the available data file types.

With deduplication, though, the data is altered substantially. Reduction rates can range from 4:1 up to 20:1 and with specific data types can even be reduced to 200:1.

Vladimir

4

Hi Vladimir,

Thanks, that makes sense!

Warm Regards,

Denis