Knowledgebase/Blog

A tale of two cities - Lincoln and Leicester

Andy Powell
posted this on Jan 24 13:51

On Wednesday and Thursday last week I travelled up to Lincoln and Leicester, meeting with Joss Winn and colleagues on Wednesday to talk about their potential use of our cloud for the JISC-funded Orbital project and for more general uses within the institution, and attending a meeting organised by the BRISSkit project on the Thursday.

Joss has already written up the Wednesday meeting so I won't say much more, other than to note that it was the first time I have demonstrated our vCloud Compute service live (using the vCloud Director user-interface) which led to a much more engaged (and informed) discussion about how our infrastructure might be used by Lincoln than if I'd only been using Powerpoint slides. I came away with a string of questions, including the following:

  • Can institutional Red Hat licences be used in the cloud?
  • What roles are available for users and how granular are they? For example, can we limit modification of vShield Edge devices to a small number of users?
  • Can we create sub-organisations within our Lincoln virtual organisation?
  • Which Linux distributions do you support?
  • Can we use IPtables in the VM OS instead of configuring the vShield Edge firewall?
I'll try and answer these in the Support section of the Knowledgebase over the next few days.

The BRISSkit event was pretty interesting, with some good presentations (including an overview, but no demo, of the BRISSkit tools - Onyx, i2b2, caTissue, REDCap and C3PR - and a rather nice outline of the benefits of cloud provision by Nick Holden) and lots of discussion about the use of cloud for biomedical research and the necessary service levels and security requirements. As a non-biomedical person, some of this discussion was well over my head. Towards the end of the day, participants were asked to comment on their willingness to use the BRISSkit tools hosted in the cloud and whether they felt them to be viable and cost effective. This was a pretty tall order, in my opinion, given that most people in the room hadn't seen a demo of the tools (in part because of networking problems on the day)... nor had they seen any actual prices.

This was a useful reminder of my findings above, i.e. that discussions around shared services only really get concrete and interesting when you can both give firm details on pricing and show what you have on offer in the flesh so to speak. Up until that point, everything can be warm and fuzzy and, in some cases, academically interesting, but no more.

That said, there were some interesting issues raised on the day and we need to think about how we respond to them (particularly if we are interested in offering community cloud services to the health sector):
  • What levels of assurance (e.g. IL2 vs. IL3 provision) are required by biomedical researchers and by the hospital trusts they work with (and what are the cost implications that derive from them)?
  • There was concern about lock-in to particular cloud providers (rightly so) and we need to think about how we re-assure people that we are not trying to lock them in to our particular cloud offering(s).
  • The business models for the BRISSkit tools are not yet clear (either individually or collectively), nor for any cloud provision of them. As I mention above, this made discussion about cost-benefit very difficult on the day beyond a rather bland, "yes, the tools look useful and I'm interested in trying them" kind of level. As a provider of infrastructure, we are clearly interested in the success of the business models for any SaaS offers running on our platform and we are keen to work with providers in whatever way we can.