[rrd-users] rrd + zfs compression = massive space savings

[Simon Hobson]

Tntteam wrote:

>So when the rrd"s are full, the gain of space will be like how many 
>% ? Any idea ?

That's not easy to predict (and will depend a little on the data 
stored), but I'd imagine you'll find the compression drops quite 
dramatically.

When you start with an empty rrd file, it will contain a lot of 
zeros/nulls and obviously that will compress very readily. When the 
rrd file is full, it will contain a lot of what might look like 
random data which might not compress much at all.

As I mentioned earlier, it will depend on the data. If you store 
'analogue' values such that all the values are different, then 
compression will be poor. But if you were storing 'discrete' values* 
then there will be better compression as there will be a lot of 
duplication.
* One example might be "number of available servers in a cluster" 
which might range from 1 to (say) 25 - but will tend to have a lot of 
repetition if all 25 are normally available.

-- 
Simon Hobson

Visit http://www.magpiesnestpublishing.co.uk/ for books by acclaimed
author Gladys Hobson. Novels - poetry - short stories - ideal as
Christmas stocking fillers. Some available as e-books.