4 comments

[ 1.9 ms ] story [ 21.2 ms ] thread
Great article!

Looking forward for the 1 billion object article as we had lots of problems with 2 billion objects inside a single bucket in Ceph Luminous.

Yes, two billion in a single bucket is a bit much, index sharding only goes to ~700M if I remember correctly.
> Least but not last, we are introducing an extended lifecycle option, bringing Ceph support options to a full five years.

I hope this doesn't mean that Red Hat will wait 2 years more than normal to offer support for nautilus.

But I'm looking forward to "Beast mode". Seems like it's time to upgrade our ceph-ansible scripts...

We support multiple versions in parallel. Currently Jewel (which has just received the 5-year extension) and Luminous, with Nautilus set for release later this year.

Beast is pretty cool. Full performance analysis coming in September!