Fresh on the heels of his talk on achieving total data center victory at Collaboration Summit in April, John Mark Walker, Gluster community leader at Red Hat will show us how to get there at the Gluster Workshop at LinuxCon Japan
on Friday, May 31 in Tokyo.
The full day of talks will focus entirely on the open source GlusterFS distributed file system and will include:
- Creating a Shared Storage Service with GlusterFS
- Converged Infrastructure: Leveraging oVirt and Gluster for Fully Virtualized Environments
- Hands on Development: Getting Started with GlusterFS translators, GluPy and libgfapi.
Here, Walker discusses the upcoming workshop, the rapid adoption of GlusterFS in Japan, version 3.4 and QEMU integration, new Gluster community initiatives and the state of enterprise storage.
What is the Gluster Community Workshop taking place at LinuxCon Japan? Why is this an important activity that you host with The Linux Foundation?
John Mark Walker: We have an amazing community of users and developers in Japan, and they responded very positively to last year's workshop. We wanted to create an event that provided this community with access to information and tools that help to make them more successful. The Linux Foundation has proved to be a great partner for the Gluster community in general and for this event in particular. Not only do they provide all of the logistical support, they also act as a gateway to open source communities in Japan.
Why do the event in Japan?
We do many workshops globally, including some that we host on our own, as well as other Linux Foundation events. We were really surprised by the turnout to last year's workshop in Japan - the Linux Foundation helped us access a burgeoning community that we didn't know the scope of. We knew there were some people using GlusterFS in Japan, but we had no idea it was so pervasive. Now that we know, we're investing more resources into the region because there's so much demand for more and better information.
We understand you'll kick off the workshop with an overview of the project and a look at the roadmap. That's pretty enticing, especially considering most open source projects don't have much in the way of a roadmap. Can you give us an early look at a couple milestones to keep our eyes out for?
Well, there are roadmaps and then there are "roadmaps" :) What we have are a list of desirable features that we would like to put into each release cycle. In that sense, we're very similar to many open source projects. We'll have a broad list of features, some discussion, and then a planning session where we identify the most likely features to make it into a release.
As for things to watch for, there are a couple. For one, GlusterFS 3.4 is currently in beta, and this is a transformative release for a couple of reasons. For one, this marks the first release where the most prominent feature was contributed by a non-core member of the engineering team: QEMU integration. For another, this particular feature greatly increases the number of recommended use cases for GlusterFS. The new block device support for QEMU/KVM bypasses the FUSE module that our GlusterFS client uses. Now you can provision virtual machines on a GlusterFS volume without the context switching overhead that comes with FUSE, and it makes a big difference performance wise. The FUSE module with the GlusterFS client is still the recommended method for scale-out NAS use cases, but the FUSE bypass using a new library, called libgfapi, will be a starting point for many non-NAS use cases, including the QEMU integration mentioned above.
Another thing to watch for is that we are in the process of greatly expanding the community charter for gluster.org
to be the center of development activity for open software-defined storage. You saw the first manifestation of that with the announcement of the Gluster Community Forge
. My magic 8 ball tells me that there might be future announcements in that same vein.
What's the state of enterprise storage and how is Gluster contributing to advancing it?
The state of enterprise storage is chaotic and rapidly changing, much like the rest of enterprise computing in general. Unlike the rest of enterprise computing, which has been in a state of flux for over 15 years, ever since Linux made inroads into enterprise workloads, this is new ground for storage.
Not so long ago, the mere idea of replacing your proprietary storage technology with an open software-defined storage solution on commodity, off-the-shelf hardware made little sense. By having many production deployments in large scale-out environments, we've demonstrated that this is entirely viable for many use cases. There are still some areas where this type of solution doesn't make sense yet, but it's early days. Right now, the storage industry is green field for open software-defined storage solutions, which is why Red Hat acquired Gluster, Inc. in the first place and created Red Hat Storage.