A few weeks ago, Yahoo’s Flickr photo service announced a big upgrade. It allows users to store up to 1TB of hi-res photos for free. I’ve been a Flickr user since 2006 and this is BIG NEWS. And as I’m writing this post, my home system is uploading over 26,000 photos to Flickr.
How can Flickr afford to do this? They couldn’t afford not to. There was a fear that Flickr had lost its mojo and needed to do something dramatic to regain its former glory.
Enter Ceph: It enabled them to do this cost-effectively. For an explanation of how Ceph “came to the rescue”, you can see this April 2015 blog from the Yahoo engineers.
As a Flickr user, I can attest to the dependability and performance of their Ceph based system.
But Yahoo is already looking at other uses of their Ceph infrastructure. In the blog, they specifically call out the need to “Optimize latency for small objects: Many use-cases such as serving thumbnails and serving during image search have small objects of the order of a few kilobytes.” So you can bet that Yahoo will be paying much closer attention to Ceph performance testing.
Today, we support Ceph testing via OpenStack Swift and Cinder, and Amazon S3. Check out this page for more details.
Sr. Director of Product Marketing