What does the future of cloud computing hold for universities and colleges? We recently sat down with public cloud providers Amazon, Atos, Google and Microsoft and university IT leaders and practitioners to discuss blockers and enablers to more widespread adoption of cloud technologies.
Cloud computing has already had a transforming effect on research, education and administration, particularly through the widespread adoption of cloud based collaboration suites such as Google Apps for Education and Microsoft Office365 through deals brokered by Jisc. For example, Office365 is now in use at over 100 UK universities.
However, Jisc’s summer 2015 cloud survey of chief information officers (CIOs) and directors of IT found that there are still a number of barriers to widespread adoption:
- 29% of the CIOs responding to our survey told us they had no plans to use public cloud to support research at their institutions
- 31% indicated they were reluctant to move business systems to the cloud
- 61% of respondents said financial aspects of cloud computing were a major concern
We thought it would be useful to bring together cloud providers and representatives from the sector for a frank discussion about these issues and what could be done about them. We were delighted that Amazon, Atos, Google and Microsoft were all able to meet with us at London’s Digital Catapult Centre in November 2015 for this ‘think tank’ on the future of cloud computing, along with IT leaders and practitioners representing a number of institutions and research facilities.
The delegates from institutions felt there were a mixture of actual and perceived cost issues around public cloud. For example, most public cloud providers charge users to download their data, so-called ‘egress charging’ – this is a cost that can be quantified as costs appear on cloud providers’ price lists. Conversely, because cloud resources are typically offered on a pay-as-you-go basis, careful planning and cost modelling is required to ensure that workloads run within their allocated budgets – hence we sometimes hear apocryphal tales of postdocs maxing out their supervisor’s credit card balance over a weekend.
Our portal onto Amazon Web Services provides some tools to assist institutions, departments and research groups in monitoring and planning for their use of cloud resources, and this was felt to be a useful development that could be built on further.
Representatives from the research community also noted that a consistent cross-research council approach to bidding for cloud resources would be very welcome. This is something that we will be exploring with the newly from Research Councils UK (RCUK) Cloud Working Group, which I have been invited to co-chair.
There was a feeling from the group that more could be done to raise awareness of the ‘shared responsibility’ model that prevails around security in cloud computing which gives responsibilities to both the cloud provider and the user.
For example, where an institution’s IT department is acting on behalf of an end user, the department may be required to manage encryption keys for data at rest in the cloud, install and maintain software installed on cloud server instances, and manage firewall settings. It was noted that some classes of data are legally required to remain within the UK, and new UK data centres announced by Amazon and Microsoft show that public cloud providers are responding to this need.
In addition to reviewing the issues noted above, the think tank meeting also considered the recommendations for R&D actions proposed in our future of cloud computing report, released in October 2015. These were:
- Cloud as a utility – we proposed that Jisc should work with the sector and public cloud providers to make it easier for institutions to switch between cloud providers and to migrate workloads between the public and private clouds, leveraging public investment in the Janet network
- App as a service – smartphones and tablets have established a culture of ‘package once, install a billion times’, but our colleges and universities still commonly package common applications separately and independently for distribution to users. This is inefficient and it diverts highly skilled IT staff from activities that could genuinely add value., but perhaps we can find a way to `package once’
- Building capability – institutional practice and experience with cloud technologies is still quite variable. We believe there is a need to come together to share our experiences of what works, building capability at institutions and identifying areas where further intervention would be helpful
Think tank delegates felt that workload and application portability were laudable goals, but that it was probably unrealistic to expect to find a general solution to these problems. Instead our efforts would be best focused on particular domains and use cases. Examples of these might include orchestrating cloud resources to provide a High Performance Computing (HPC) platform, ‘canned’ workflows for particular scientific computing tasks such as BLAST for genomics, and packaging Windows applications for remote delivery.
There was significant interest in technologies such as OpenStack that might form the basis of private institutional and community clouds, and potentially give institutions a pathway to public cloud services where appropriate. Delegates also felt that container technologies such as Docker had huge potential as the ‘Rosetta Stone’ for application portability.
Containers provide a convenient way to package up an application and all its dependencies, and then run it in isolation from other applications, without incurring the performance penalties of traditional hypervisor based operating system virtualization. The recent formation of the Open Container Initiative with widespread industry support suggests that Docker will evolve to become the de jure standard in this area.
Amid the think tank roundtable discussions, one question was repeatedly asked – what will an IT department’s staff profile look like in five to ten years? Right now a significant proportion of staff hours at most institutions’ IT departments are devoted to managing hardware and packaging and distributing software.
Delegates felt that moving to the cloud created huge opportunities for the IT department to become a true business partner and enabler for institutions’ key activities around teaching and learning and research. Delegates also felt that there was significant potential for additional Jisc brokerage activity around cloud products and services.
The future of cloud computing think tank was organised as part of our horizon scan activity. This aims to identify and discuss major trends in digital technology affecting higher education, further education and skills, and feed into our strategy setting process. In this case we will be using the results from the think tank to inform our cloud strategy, which is currently under development.