Hi Folks,
Earlier this spring I did some work looking at space amplification with
bluestore under different workloads and min_alloc sizes for RBD, Rados,
and RGW. I was recently asked about metadata overhead in bluestore
again, and I realized I don't think I ever shared the results with the list.
The gist of it is that metadata overhead is relatively low if your
objects are large. Especially if you are dealing with small RGW writes
however, you can start to see fairly significant metadata overhead (~48%
for 4K RGW objects). These results also showcase how the min alloc size
affects space amplification for small objects. For very small objects
with a large min alloc size (Say 64K for bluestore on HDD) the
amplification can be very significant. So long as your objects are
typically larger than the min alloc size, this isn't really an issue though.
You can see the results here:
https://drive.google.com/file/d/1Ews2WR-y5k3TMToAm0ZDsm7Gf_fwvyFw/view?usp=sharing
Thanks,
Mark
--
To unsubscribe from this list: send the line "unsubscribe ceph-devel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
