Friday, 29 October 2010

Resourcing academic computing

Have just been to a rather interesting presentation on the ARCS data fabric.

The ARCS data fabric is in essence an initiative to build a shared storage cloud in Australia for use by researchers and also to combine this with a grid based execution environment.

I have previously written about how people tend to assemble their own toolkits of resources, and how this has some odd effects, such as wikidot becoming a wiki provider of last resort in academia. It's also the case that any toolkit of resources should include some offline storage for key documents if for no other reason that hard drives break and computers get stolen. And often an extra benefit of remote storage is that they have a permissions model meaning that you can share some of the content of your remote store with friends and colleagues, some with the world and keep some private - in effect owner:group:world. And one example of such a service is windows live's skydrive, which is bundled with windows live accounts, and which is free at the point of delivery.

Now interesting things about the ARCS data fabric is that it uses webdav and provides very similar sharing functionality to Windows Live skydrive (and the same amount of storage as default) and is currently free at the point of delivery.

Unlike the ARCS service Skydrive does not come with a desktop connector but Gladinet will sell you one that will link to a whole range of storage providers, including webdav hosts and windows live, allowing cross mounts - which for example allows you to move content from Google Docs to Windows Live.

One of the great debates we always have in university computing is whether to outsource email, and if we do, whether to choose Microsoft or Google.

Ignoring theology, the differentiator has always been that Google provides better tools in the form of Google Apps including shared document editing, and Microsoft provides better storage in the form of skydrive. It's also true to say that Microsoft is more windows oriented, but with the new Office web applications they are becoming much more agnostic about such things and that their web apps are now as functional and environment-agnostic as Google's

So, purely for the sake of argument, let us say we outsource email to Microsoft, and use the money we save to licence an appropriate connector to mount skydrive on the desktop. Now we know we don't save a lot of money getting rid of student email as we know students increasingly self outsource and use a webmail service for email. If we don't also outsource staff email we end up having to provide almost as much infrastructure as before.

So there is no great win in only outsourcing student email. But remember that going to Microsoft gets you skydrive. And this could be a radical opportunity, possibly too radical, and decide to no longer provide student filestore and tell them to use skydrive.

Now, we do save money by not providing student filestore, providing performance is at least as good as any existing student filestore, and that we trust bigM not to lose any data. Student data lives in the cloud and ideally is accessible from any location on any platform via a browser at a minimum.

Now the endpoint of this is we end up providing a service very much like the ARCS data fabric to students and by implication, all members of the university, and even these days 25GB is a reasonable amount of online storage.

So, in this scenario, do we see Microsoft as a suitable provider of storage for work in progress researchers and do we then see initiatives such as the ARCS data fabric turned into providers of specialist storage for either large datasets or to get around legal/jurisdiction related problems with sensitive data?

And does that mean we increasingly see a landscape of outsourced filestore?

No comments: