Red Hat has released version 1.2 of Inktank Ceph Enterprise, bringing the benefits of erasure coding and cache tiering to the open source software used for managing ‘hot’ and ‘cold’ data.

Ceph, an open source file-and-object store technology, was born out of the University of Santa Cruise with a grant by the US Department of Energy.

It was commercialized by Inktank in 2012. It originally wrapped services and support around the software for commercial users before creating an enterprise version.

Red Hat purchased Inktank in April this year for US$175m with the idea that it would be complementary to its own Gluster distributed storage technology.

Red Hat director of product marketing for storage and big data Ross Turk, who moved over with Inktank said the company is only in the early stages of its integration with Red Hat, and trying to work out how to tie Ceph and Gluster together.

“There are a lot of dust clouds that need to settle when two companies merge,” Turk said.

The new release of Enterprise Ceph 1.2 has been very much an Inktank project.

“We were told to get Ceph 1.2 out the door first but now we will look into pushing through the Red Hat engineering process,” Turk said.

“This will see us take a really good look at both Ceph and Gluster. The more we look at these, however, the more we come to the conclusion that they are very different architecturally. Object store is the exact opposite in Gluster, and they both do two very different things – the way they managed themselves, the limitations of each architecture and their strengths.

“It seems like it might be a simple answer at first – a use for this case, and a different one for that – but as we dig in we find each architecture has quite a bit to offer, and it will take quite a bit of time to figure out how these products can be explained to the market together.”

For this first release under Red Hat ownership, 1.2 brings erasure code in to the mix for the archiving of cold storage – erasure code is a forward error correction code for binary erasure channel. It can provide redundancy by breaking up objects into smaller fragments which can be stored in different parts of the overall storage architecture.

Ceph allows multiple pools of storage to be contained in the same storage cluster.

New cache teiring allows for hot data to be moved on to high performance media when it becomes active, and cold data to be moved into lower performance media when no longer in use, brining about further efficiencies.

Inktank has also made changes to its Calamari management platform, allowing for autonomic distributed object store storage cluster, which allows for the management of both devices and pools of storage.

And of course there is also further integration with Red Hat Enterprise Linux OpenStack Platform 4 and 5.

Turk said to date, the benefits of the commercial version of Ceph have appealed customers using Ubuntu. But in the last six months demand has been picking up from Red Hat Linux customers.

“We are also seeing a lot of interest from people using OpenStack,” Turk said.

“A lot still think storage is a hardware concern, but it is increasingly becoming a software concern and this is driving demand.”

What is really driving demand, however, is the growing complexity of managing storage environments.

“You see a lot of technology out there that increases the amount of manual work when the cluster grows or shrinks, simply for reallocation. This doesn’t happen with Ceph, Ceph architecture is different from these storage architectures as it looks more peer to peer,” Turk said.

Ceph also uses an algorithm called CRUSH which calculates where objects are and makes them tuneable to policy management.

“This means when you define a policy for your cluster, your failure domains, your processor, server. Ceph makes everything aware of where it needs to be and you can get the same calculations and results regardless of where the data is,” Turk said.