ICT strategy: a strawman proposal Prof. Mike Giles giles@comlab.ox.ac.uk Abstract This is my first draft of a personal strawman proposal and an accompanying questionnaire to get feedback from academics. 1 Introduction An IT strategy is a hard thing to write. On the one hand it runs the risk of being too vague, making general declarations about the importance of IT but leaving the details to be decided upon later. On the other hand, if it is too specific it runs the risk of becoming obsolete far too quickly due to advances in IT technology. A compromise therefore may be to have a periodic review of overarching principles (perhaps every five years) accompanied by a technical plan which is updated yearly. Each would require input from the university community, and there would be debate about both the general principles and the specifics of how they are to be realised through the technical plan. Following this approach, I have divided the remainder of this document into two sections, one on general principles, and one on more detailed ideas of how the principles may be realised in practice. The aim is that the general principles should remain invariant over a five year timescale, whereas the technical details are likely to change over that time. 2 General principles 2.1 Subsidiarity, accountability and communication Perhaps the largest strategic question concerning IT provision within the collegiate university is the appropriate level at which different services should be provided. Which services should be provided locally by the department or college, and which should be provided centrally by OUCS, OULS or MIS? The benefits of local provision are: good responsiveness to local user needs good interaction with users in setting priorities, deciding on cost/service tradeoffs 1
while the potential drawbacks are: increased staff costs through duplication of effort lack of provision in specific areas (e.g. database development) where the need is insufficient to justify employing a full-time person Correspondingly, the potential benefits of central provision are: reduced staff costs through economies of scale in providing a single service for the whole university clear specification of standards ensuring smooth operation across the entire university ability to provide specialist help in niche areas while the potential drawbacks are: poor responsiveness to the needs of end-users increased costs through the provision of services for which there is no real demand Q: for each of the above benefits/costs, do you agree/disagree? Are there other benefits/costs which should be highlighted? In general, it is desirable that the appropriate level of provision (local or central) for different services is decided by consensus. Colleges of course retain the right to determine their own IT strategy, but are likely to see the financial and technical benefits from central provision of some key services. Departments should also have the right to determine their own IT strategies to a large extent, but should be prepared to justify them to divisional or central university oversight. Accountability and communication are key aspects of an effective and cost-efficient provision of IT services, with the goal being to avoid some of the potential costs outlined above. In the case of services provided locally within departments, there should be: webpages documenting the service provision and the level of support provided for those services which duplicate central services, an explanation or justification for the duplication The former is to provide information for users within the department, and to encourage communication and the sharing of knowledge and best practice between departments. The latter is to ensure that departments do not needlessly incur unnecessary costs, but may often serve the vital role of providing information to the central provider on the deficiencies in their service, Q: do you agree that departments should have the right to opt out of central services? do you think they should then have to justify this? In the case of services provided centrally, there should be: 2
webpages documenting the service provision and the level of support provided an effective means for obtaining feedback from end-users to improve the service provided an appropriate means for considering the cost-effectiveness of the service provision, so that costs are not allowed to escalate unreasonably in responding to specific users requests for service improvements There will always be a trade-off between the quality of service provided, and its cost. That tradeoff is much more easily managed within a relatively small unit such as a department than it is with a centrally-provided service. Thus, extra effort needs to be taken with central services to ensure the appropriate level of service provision. Q: do you think the current feedback mechanisms for OUCS services are adequate? what changes would you like to see? With regard to IT provision within colleges for undergraduate and graduate students, and also academics who work primarily within their room in college, there should be be a documented policy on the support provided by the college, the relevant department and the central university, so that all parties are clear as to their rights and responsibilities. Q: do you think there are specific problems at present concerning who is responsible for the IT needs of students and college-based academics? Finally, it is very important to have good communication between all of the IT staff in colleges, departments and the central university. This ensures a good understanding of each others needs and challenges, and efficient sharing of expertise and best practice to avoid unnecessary duplication of effort. Q: do you think the existing formal and informal links work well? do you have suggestions as to how they can be strengthened? 2.2 Oversight of major ICT projects There are numerous examples in the past of major ICT projects within universities as well as within government and industry which have failed to meet their objectives, have failed to provide a satisfactory service to the end-users, or have done so at a vastly increased cost. To avoid this requires good communication and accountability. Any large ICT project within the central university should have an oversight committee consisting of users, who are responsible for ensuring that the users needs are properly understood and addressed appropriate experts, who can scrutinise the plans and implementation to independently question their appropriateness, senior management, to ensure that the concerns of the users and experts are properly accounted for, and to take action if they are not 3
This, together with a lengthy period of consultation beforehand to gather the users requirements, and a consultation afterwards to determine the users satisfaction with the delivered system, should reduce the risk of major difficulties in key ICT projects which are often crucial to the smooth running of the university. Q: do you agree with the importance of an appropriate oversight committee for major ICT projects? do you have other suggestions on how best to avoid major problems with large ICT projects? 2.3 Some technological considerations Specific proposals about which core services should be provided centrally are made later in the technical plan. However, one strategic policy decision should be that, as far as possible, such services should be based on platform-neutral open standards. This means that the services can be accessed by users with computers running any major operating system (Windows, OS X and various kinds of Linux and Unix) and should be based on open standards, rather than proprietary specifications, to avoid users being tied to a specific vendor s software. This increases flexibility and reduces software costs. Q: do you agree with the importance of services being based on platform-neutral open standards, as far as possible? Another element of the strategic plan is that university should keep watch on the development of new ICT technologies which may have a significant impact in the future, running pilot studies where appropriate to reduce the uncertainty and risk often associated with new technology before deciding whether, when and how best to deploy it. 3 Technical plan 3.1 Existing core services There are a number of existing core services provided centrally. They are listed here grouped according to the organisation which provides the service. OUCS: university network, including firewall to external JANET connection Herald email service and associated email virus scanning and junk mail filtering HFS backup service dialin and VPN services software courses PC hardware maintenance system OUCS Shop 4
site-licensed software Weblearn virtual learning environment Q: how do you rank these services on a scale of 1 5? do you have any particular comments to make? Some suggested changes to existing OUCS policies: Herald email service should have larger quotas (e.g. 2GB for academics instead of current 100MB limit) which increase in line with technology (40% increase per annum?) The HFS backup should be universal, available to all staff and students in the university, from any location (including home and student lodgings not connected to the university network). It should also have expended capacity to be able to support all departments and colleges VPN service should perhaps be migrated to OpenVPN which is considered to be more secure and less restrictive than the current proprietary solution from Cisco Q: do you have any other changes you would like to see? OULS: web-based access to bibliographic databases web-based access to journals, books, etc technical support for departmental and college libraries a new data repository for archival storage of papers, pre-prints, etc. Q: how do you rank these services on a scale of 1 5? do you have any particular comments to make? Some suggested changes to existing OULS policies: Integration into Google Scholar and other search engines to simplify access to electronic materials Q: do you have any other changes you would like to see? MIS: ISIDORE OSIRIS technical support for departmental and college administrators Q: how do you rank these services on a scale of 1 5? do you have any particular comments to make? 5
3.2 New core services There are a number of new core services which are either under development or proposed for development, if there is sufficient demand: universal provision of personal calendars, addressbook and other information, with secure access and sharing a universal system to manage tutorial provision (including timetabling, submission of tutorial work, submission of term reports, and financial accounting) a universally accessible shared filestore, specifically to enable collaboration between people in different departments, colleges or at home. wireless network access and VoIP (both discussed later under New Technologies) shared-memory and distributed-memory supercomputing facilities (to be provided by the Oxford Supercomputing Centre) cycle-scavenging for trivially-parallel applications (to be provided by OxGrid) support for Grid-based computing (to be provided by the Interdisciplinary eresearch Centre) Q: do you have any comments on these proposed new services? are there other new core services you d like to see added to the list? 3.3 ECE: Enhanced Computing Environment The CCE (Common Computing Environment) is an initiative to develop a common Windows XP platform (hardware plus software) for use in MIS, OULS and OUCS. The idea is to gain from economies of scale in the procurement and deployment (in particular the latter) of a large number of systems to address the similar needs of the three organisations. The project should be viewed as a pilot, and it will be very interesting the extent to which it turns out that the three organisations do have similar needs, and the extent to which there are significant savings from combining them in this way. If it is successful, then the intention is to offer an enhanced version (the ECE) to departments and colleges. In essence, this is another central service, in which the end-user pays a fee and the central ECE team deliver a combined hardware/software service. As a matter of ICT policy, this will not be forced on any academic department (or, of course, the colleges which are at all times free to decide their own IT policies). It will only be offered, and it is then up to the department to assess the possible benefits. If there is significant takeup, then the intention is to look at alternative platforms such as some flavour of Linux. The choice of Windows XP for the initial platform reflects the needs and wishes of the lead three organisations, and is not intended as any form of overall policy decision. 6
The precise hardware/software specifications of the CCE have yet to be finalised. When they are, they will be available on the appropriate university webpages. Two departments and one college are being consulted in drawing up the specifications to ensure that no decisions made at this stage make the project completely unsuitable to departments and colleges at a later date. Q: do you have any comments on the CCE plans and the future ECE possibilities which would be helpful in future planning? 3.4 New technologies Wireless networking technology is becoming mature and is used in various parts of the university. However, widespread deployment may have significant costs because of the very limited range of wireless routers, and there are competing technologies such as 3G/4G mobile phone technology. The university should review the situation to: assess user needs (e.g. how many users? at what locations? bandwidth requirements?) review relative benefits and costs of different technical solutions investigate opportunities for major donations and/or special deals from major telecommunications companies Q: do you agree with these terms of reference? If not, what would you add/delete? Telecommunications is rapidly moving to VoIP (Voice over IP). The university should review its telecommunications service to: investigate the cost and potential savings of VoIP consider the implications of VoIP for the future development of the university network consider the feasibility of a pilot study (e.g. for remote buildings, or for student bedrooms) decide whether to prepare a plan for widespread implementation Q: do you agree with these terms of reference? If not, what would you add/delete? Although wireless and VoIP have been treated above as separate items, it may be that for either technical or commercial reasons they should be combined into a single item. Q: are there any other major new technologies which you think the university should be tracking at present? 7