Next session was on Cloud Computing, three different speakers.
First was from the University of Newcastle and was about the use of cloud in Research.
Large companies like Google, Amazon, Microsoft have huge data centre to run their core services. Now these are being opened up and they're giving others access on a pay per use or pay as you go model. It provides better scalability, particularly being able to cope with "bursting patters" common in research and the peaks and troughs of term time working. It also allows organisations to move capital expenditure to recurrent as it's up to the cloud provider to refresh machines etc.
Problem with clouds is it's a low base provision, if your not an IT expert it can difficult to use.
Researchers need systems to be scalable, reliable, secure. Takes skills beyond most researchers. In Newcastle, they designed a platform which sits on the low level infrastructure which cloud providers provide.
Researchers want 4 common things:
Secure data storage
Ability to analyse data
Share in controlled way
Plus, everything should work through a web browser.
Newcastle's platform is eScience central
It has a Web 2.0 look and feel with a blogging system. The security is based around social networking, as for young researchers it's a way of life. Can connect with each other, form groups etc. Security built round these groups. It has versioning and metadata in the form of tagging or XML.
Data can be analysed using blocks of code, some are provided, others can be uploaded, and a workflow system. Runs in cloud.
Now starting to use it for teaching. Because browser based can be used anywhere on anything. Teachers can set up exercises, and remotely view students work.
Support tech transfer, supports industry collaboration.
Clouds have potential to transform research, teaching and technology transfer and it's not just about cost reduction
But, moving to cloud infrastructure doesn't help the vast majority of potential users who lack deep IT skills
High level platforms are needed which handle storage and processing, scalability, dependability, security and audit
Insulates users from cloud vendor lock in. Can deploy on intranet eg medical data as well as cloud. Runs on azure, amazon, google etc
Next up was Phil Richards from Loughborough talking about using the cloud to make cost savings. Used example of HP consolidation project. They created a private cloud, reduced data centres from 85 to 6 and applications from 6000 to 3000. Had to invest $100ms to create network, equivalent of our JANET. But, they achieved savings of $1b per annum in revenue.
Loughborough have done something on smaller scale as their old old data centre needed refurbishing. Couldn't afford it so working on cloud based solution with Logicalis based on two mini pods.
In terms of HE, the killer app is not going to be research tools or shared administrative applications but the provision of cheap virtual servers or Infrastructure as a Service via hybrid clouds.
We are lucky to have a network already in place and ready to provide this service, ie JANET, which puts us in front of private sector and rest of public sector. Also JANET now acting as broker. If we get good commercial deals there's the potential for big savings.
Final speaker was Henry Hughes from JANET.
Drivers for moving to cloud. Universities being told that we're not taking advantage of savings, we're operating in outdated way, big cuts on the way.
Focus on cost reduction. Data centres can use >10% of power for an entire campus.
Carbon reduction commitment.
Real estate becoming more valuable for other activities
Some problems provisioning power in some areas
Government cloud developing the G cloud.
Is it that easy to move to Cloud? No
We often have customised in house systems , integrated with other systems
Total cost of ownership often not known, difficult to build business case, elements outside IT eg finance
Costing of IT services difficult, three things have to be taken into account - people, processes, technology
Making reliable estimates of cost savings difficult
Concerns over data security and privacy
Uncertainly over data location and legal risks
Uncertainties with regard to changes to security
Risks have to be considered, eg
Compliance eg DP, patriot act, safe harbor
Multiple tenancy, ie shared storage, provider being responsible for security,
All down to risk assessment, flexibility and value have to be balanced against the security risk
EDIT The video of this session is now available here. You need to install Silverlight (there is a prompt to do this), and using the i button on the bottom left of the presentation, select Parallel session 1, Cloud solutions.
- Posted using BlogPress from my iPad