Improve compression ratio in disk tables

We recently upgraded from version 6.8 to 7.6, but we noticed that compression ratio is still the same on disk tables.
Are there any hints to improving compression ratio in disk tables?
Do you expect to eliminate the space occupied by null columns in memory and disk tables on any upcoming release ? If yes, in which release ?

Great question Nabil! I’ll escalate this internally and respond shortly. Thanks, in advance, for your patience. :pray:

We haven’t changed columnstore compression for a while. There has not been demand for that – it’s pretty good as is. Unlimited Storage Databases do lz4 compress the already-compressed blobs before putting data on remote object storage. But the data does not get that LZ4 compression on local storage in the object cache.

Rowstore sparse compression does compress out nulls.

Nulls will be compressed in columnstore by regular compression algorithms.

1 Like

Thanks Eric, Maria. Will run some tests to evaluate whether performance is impacted when using rowstore sparse compression.

1 Like