OUCS 5-Year Plan: Major Capital Investments
The major capital investments that OUCS will need to cover fall under the area of supporting the University's infrastructure, as outlined in the Corporate Plan in VI. What follows is a considered list of the major capital investment needed by Oxford to maintain its IT and Telecoms networks, and to expand them to meet the demands of a word-class University. We are already falling behind our competitors in areas such as wireless networking, institutional filestores, and repositories; and our network backbone will need major investment over the next few years.
The backbone switching structure was upgraded to 10 Gbs in 2004. This will provide capacity until the next step-wise upgrade. On past experience this can be expected towards the end of the 5-year window. OUCS envisages that £1M should be budgeted for this. However, one area of uncertainty is the security aspects of a higher-speed network, and adequate control may depend on developments (such as high-speed firewalls) that cannot yet be predicted. Target date: 08/09. Estimated cost: £1m.
The network infrastructure is based around ducts and fibres installed in 1984. Although this continues to serve us well, the time is now long overdue to review cable provision to increase capacity and replace outdated cables. The issues of resilience should also be reconsidered. Although resilience has always been a prime consideration, a new risk-assessment, including the increasing demands of teaching, research and administration, may require extra resilience to be provided, including duplication of switching equipment and cable routes. This should be carried out in the next three years, and it is suggested that £0.9M should be budgeted. Target dates: 07-10. Estimated cost: £0.9m.
Roving access must be high on the agenda of the ICT Strategy Group. In some cases, particularly colleges, wireless access (conforming to the model already established) may be expected to come about incrementally after the foundations are laid by FroDo. Central areas, for example libraries, may be more problematic. A bid in 2005 for funds for provision of wireless access in 60 main University buildings was turned down, but this must be revisited. The layout of Oxford (and the construction of many buildings) means this is an expensive process. To equip 60 main buildings, budgetary costs are £1.5M capital. Recurrent equipment costs may be estimated at £200k. Running this network would require approximately two FTE. However, it should be noted that OUCS has received funding (55k pa) to expand the Network Support team by 1 FTE from 07 for three years. This could be directed to this project (hence the variance in cost noted below with the ICT Strategy). Increased use of this technology will open up many other areas for exploration, and the opportunity to provide services directed at PDA and other mobile users. We will also be monitoring any development on a city-wide basis by the City Council who may be looking to partnerships with commercial suppliers. However, previous discussions indicate that such initiatives would not be looking to making sure wireless was available within University buildings or areas, and would focus on tourist locations.Target dates: 07-10. Estimated costs: £1.5m hardware, £150k recurrent for 3 years; total cost: c. £2m.
This can be expected to be introduced in the next 2-3 years. This will present major technical challenges (throughout the University), but the capability is already there in the switching equipment, and no significant capital or recurrent expenditure is anticipated.
The Telephone Infrastructure is based on aging analogue technology, but at the moment is extremely reliable and fully maintainable. The next five years must see a change to VoIP technology, driven by demand for new services. Minor pilots will be carried out, but it is expected that the first major installation will be for the development of the Radcliffe Infirmary site. Funding for this will come as part of the RI site costs. However, a major transfer of the University telephone infrastructure to VoIP may be expected towards the end of the 5-year period. A migration strategy will have to be planned, and until this is done costs and timescales must be very uncertain. Target date: 2009. Estimated cost: £3m.
OUCS has been funded to install a separate ‘secure’ backbone network that will provide the basic IP-based switching structure for the telephone network, and will be available for building-management and security functions. These will be arranged as needed by the Estates Department, and funded from the building projects needing the facility.
SuperJanet 5 is expected to be installed in 2006, and it is hoped that Oxford will have a 10Gbs connection. Costs of this will be met from OUCS recurrent funding. The installation of FroDo gives an opportunity to consider again whether a separate commercial link should be provided. Some colleges are already using such a connection for conference use, and there has been discussion on whether some of the legal burdens on the University could be eased by using an ISP for residential access. FroDo would enable a central service to be provided for all colleges — it is not expected that this would come from University funds.
OUCS makes continual investment in central mail facilities to keep up with ever-increasing demand, both in the numbers and volume of email. A major investment has recently been made (from OUCS reserves) to provide a resilient and expandable fileserver for the Herald cluster, which was brought into service early in 2006. Funding has also been received in-year to boost the quota to 1Gb per user by autumn 07. If the IT Strategy leads to greater use of the central mailstore, further capacity will be required in the future, and extra back-up facilities provided. A further review of expansion should take place in 09/10.
There are growing requirements for user identities to be handled in a strategic manner, so that departments, faculties, colleges, and central services have single sources of information to cover authentication and authorisation needs. This has to be co-ordinated by BSP, but OUCS will need to do a considerable amount of work in the background (e.g. providing robust reliable servers). Target dates: 08-09. Estimated costs: £220k.
OUCS has begun an analysis of existing groupware packages in response to growing user demand. Groupware packages combine a series of systems, such as email, calendaring, scheduling, and so on, into a single environment. This includes systems such as Microsoft Exchange. If our analysis indicates that such a system would be required then investment would be needed in hardware, staff, and possibly software licences. Target dates: 07-10. Estimated costs: £20k hardware, £500k pa licences, £100k staff pa; total - £1.82m (for three years plus recurrent costs thereafter).
OUCS (through RTS) will continue to play a coordinating role in the development of collaborative research environments and should lead its development. Currently, however, VREs are still experimental and domain-specific . By 2008 OUCS should be clearer about what common VRE interfaces and tools ('services') will be managed centrally and what needs to be managed by research groups. OUCS will continue its coordinating role and seek funding to scope the institutional VRE services at the earliest opportunity. Target dates: 08-10. Estimated costs: £50k hardware, £100k recurrent costs; total (over three years) £350k.
The HFS Service must constantly expand to keep up with demand, and to cater for new and developing areas. Amongst areas that may need considering in the future are undergraduate backup, PDAs and other mobile devices, and system images. The last major upgrade (with HEFC funding) was completed in 2003. Extra storage capacity has recently been purchased from OUCS recurrent funding taking the current configuration to capacity, and was installed in early 2006. Also during 2006, OUCS drew on reserves to fund the more pressing major upgrades, replacing the ageing p-Series processors and the older tape drives, as well as replacing the ageing tape media. Further updates should be planned during 2007 to upgrade the 3592 generation-1 tape drives to generation-2 and to provide additional disk storage servers. In 2008 it will be important to replace the p-Series servers installed in 2003. A budgetary cost for these three items is £770k in total (Project: HFS: update servers/disk storage/tape drives). In the slightly longer term a review is needed of on-site and off-site storage, with an intention of installing a second tape library at a separate location. This, apart from giving greater security, will relieve capacity bottlenecks by allowing duplicate copies to be made off the main library. A budgetary cost for this is £0.5M (Project: HFS: offsite library). Obsolescence of the 10-year old automated tape library at OUCS has begun (i.e, some parts are no longer being sold) and its replacement will become necessary around the end of the 5-year period. It is difficult to predict costs so far in advance in a changing market, but on present costs about £550k will be needed (Project: HFS: replace central tape library). Target dates: 07-11. Estimated costs: £ 1.8m.
An objective over the next three-four years should be to provide a more unified approach to support of backup, using a hierarchy or cluster of TSM servers and tape library devices, some of which can be located in user departments, giving them more direct control of local backup while still using the deep-storage facilities of the HFS. This could also lead to a greater emphasis on facilitating faster recovery of vital systems. It is expected that the ICT Strategy will address this general question. We envisage that this should be done on a cost recovery basis.
The areas of filestore and repository must be a priority for consideration in the ICT Strategy because (a) the amount of data that the University holds is growing exponentially, and there is at present no framework for managing that data reliably and securely; (b) there are areas of new and pending legislation which will impose obligations on the University, and where it is necessary for it to be able to demonstrate compliance; (c) increasingly the University's assets, in particular its library material and archives, will be held (sometimes solely) in digital form. Any policy must address two elements: a common tree structure so that data anywhere in the University could be accessed by anyone (with appropriate authorisation); and a central fileserver architecture providing a range of facilities, including a large-scale, high-performance repository, with secure backup and off-line facilities. Clearly a lot of this would be outside the remit of OUCS. Without guidance on the extent to which this might be addressed cost cannot be determined, however as an indication, a 100TB SAN-based system (one of several options) should be expected to cost about £0.8M for each of two phases, and require at least 2 FTE to support it. Target dates: 07-10. Estimated costs: £2.7m.
Unless OUCS relocates the pressure on existing user areas will have to be alleviated. Although three of our lecture rooms have been modernised recently (one through the HEFCE E-learning fund), the last one will need to be worked on soon, and possibly a fifth lecture room may need to be built. In addition our help centre facilities will need to be developed in keeping with our review of the front of house services. Work on both of these should commence in 07. Target dates: 07-09. Estimated costs: £600k.
OUCS is currently suffering from sever limitations on space for staff accommodation and storage. The University Estates Office have identified two possible areas we could consider which would add value to the overall building — namely refurbishment of the basements and installation of a service lift to ease access to large (unused) storage areas. Target dates: 07-08. Estimated costs: £425k.
The OUCS machine room is currently running at capacity and there is no room to accommodate extra servers. Furthermore, at certain periods there is a risk to key services due to the pressure on power and air-conditioning. As has also been highlighted, the single point of failure with OUCS's machine room implies considerable risks for key services within the University. Target date: 08/09-10. Estimated costs: £17m (N.B. this figure was drawn from the ICT Strategy which was considering a University-wide machine room available to all departments, and assumes that much of this could be cost recovery. OUCS's machine room needs should be considered separately — and in turn would be considerably lower than £17m).
The following lists the major capital projects identified above, and presents a slight modification of the table in the Draft ICT Strategic Plan (Appendix A). N.B. This does not cover all new services, expansion to existing services, or the need to address understaffing of core services, as some of these may simply be recurrent costs.