Tuesday, March 13, 2012

Riding the hype curve and shooting for the moon at OGF34

At ISGC2012 last week, we were introduced to Gartner’s hype curve and its application to new technology. With any new widget, gadget or pan-European infrastructure, you start off by riding rapidly up the hype curve, coast over the peak of inflated expectations, plunge down into the trough of disillusionment, creep painfully back up the slope of enlightenment and hopefully emerge triumphant out onto the plateau of productivity. There was much discussion in the bar afterwards about where exactly on this curve you might find grid and cloud computing.

Chair of the OGF Board, Steven Newhouse started off with the Gartner hype curve in his talk yesterday, and told us that cloud is a disruptive technology across all sectors – but is it really easier to use than grid? We are now at the point where isolated institutional clouds are starting to emerge, in the same way as we used to have isolated commodity clusters, subsequently linked together through grid. Just as for clusters, federation is the way forward for isolated clouds. “This is a grid federation problem,” said Newhouse. “We know how to do this!”

OGF can bring its grid experiences to the cloud arena through an active community, who have a decade of expertise in best practices and standards. Hence OGF’s contribution to new technical areas such as infrastructure cloud management, through standards including OCCI. Clouds are a novel resource and need to be federated so they can be used efficiently. “We can adapt grid technology to federate clouds,” summarised Newhouse, “and in future this will be relevant across many public sector areas.”

From riding the curve, to shooting for the moon in the next session. Josh Howlett of Janet introduced us to Project Moonshot. This is a Janet-led initiative, in partnership with GÉANT and others to develop a single unifying technology for extending federated identity to a broad range of non-Web services. These include cloud infrastructures, HPC and grid infrastructures, as well as more common-or-garden services such as mail, file stores, remote access and instant messaging.

Janet for example had three services: Identity federation based on SAML (Security Assertion Markup Language), Eduroam, the research and education community access service and certificates based on X.509, the public key infrastructure.

Each of these services were governed by different institutional policies, which led to a less than seamless customer experience, even though conceptually they seemed to be doing the same thing. Moonshot’s goal was to offer a single solution for trusted identity for any kind of application.

Howlett discussed several use cases for Moonshot, including grid computing at the Science and Technology Facilities Council, STFC. The existing X.509 authentication is often too complex for users and the goal here was to simplify access when certificates were not really needed. A further use case was with Diamond Light Source, the synchrotron radiation source at the Rutherford Lab, aiming to give scientists access to the Diamond consoles both locally and remotely. A further application has been developed with Cancer Research UK, which will help different institutes to share data and documents that would otherwise only have been available internally.

Moonshot want their solution to be scalable, up to literally millions of users. It builds on eduroam technologies, EAP (strong mutual authentication) and RADIUS (federation between domains) and to this adds SAML for rich authorisation semantics. The team has tested many common scenarios, such as Firefox/Apache, IE/IIS, MyProxy client/MyProxy Server and lots of others, generally with success. So in future the user’s experience should be better and simpler, and the costs of providing services should be lower for the institutions providing them. Sounds like a reasonably clear shot at the moon to me.

No comments: