White dot for spacing only
The Dice Project

Final report for project 303: AT 3/4/5 decant to Forrest Hill,
and also project 304: AT 6/7/8 decant to Wilkie Building and South College Street

gdmr, July 2015

One pseudo-project (303) was initially set up to track the work involved in the decant of Appleton Tower levels 3, 4 and 5 to Forrest Hill while the recladding is done. A separate project (304) was set up to track the move of levels 6, 7 and 8 to Wilkie building and South College Street. In fact, though, due to the overlap of people involved both projects were undertaken as one, and this "final report" covers both together; and in hindsight one project to cover them all would have been more straightforward.

NOTE: We only account here for the CO time spent in the design, planning, setup and installation. There was also a considerable amount of admin-staff, technician and CSO time involved, only some of which is noted below. Neither do we account for the productive time lost by the CO/CSO team as a result of their decant, as there are separate projects to cover this. The estimates below are for time actually expended on the projects, not elapsed time which was well over a year from start to finish.

A separate project will be set up to cover our expected return to Appleton Tower in 2016.

Design Team Meetings

Key to the success of these projects, at least from our point of view, was to be engaged early and completely in the design process for all of the buildings. In the event, both Wilkie and South College Street were relatively straightforward: there were only a couple of design-team meetings, specifications could be mostly sorted out through email, and a handful of building walkrounds sufficed.

Forrest Hill was a completely different kettle of fish. The building needed substantial refurbishment, to be done in a way which matched our teaching requirements as well as possible, while remaining (reasonably) within E&B's budget.

Time estimate for this stage: 3 weeks of actual project time, or around 15% FTE over the period.

Design and Procure Network

As the expectation was that we would only be in the buildings for a relatively short time, the decision was made to recycle many of our recently-replaced Forum switches rather than buying new ones. (Aside: this may have some implications for E&B once we move out and IS start to kit out the buildings. Whether or not we leave them our old switches, patched in and ready to go, is still up for discussion.) We were also able to reuse the 10GbaseLR X2 modules which were bought for use the last time we decanted out of Appleton Tower to Forrest Hill.

We did, however, buy a couple of new switches for Wilkie, as we didn't have enough leftover X2 10GbaseLR modules and it simply didn't make sense to buy something which wouldn't be reused later. These were bought from our usual framework supplier, and the cost recovered from E&B. This was done in October 2014 (RT#69041) to allow for the usual lead times from order to delivery, and to give us plenty of time to test and set up all of the 34 switches required.

Once the port layout was finally frozen and the numbers on each floor of each building set, the actual design of the network (by George) was straightforward, based on our usual principles of redundant links, avoiding single points of failure and patching in everything everywhere. Diagrams for Forrest Hill are here and for Wilkie Building here.

Project time estimate for this stage: 1/2 week.

Set Up Switches

Once the additional hardware had arrived, we installed the AT core modules and patched them through to the Forum. This was done to make setup of the edge switches more convenient, and also gave us a way to test the basic design. George installed the initial 1wb? and 0fh? switches, these being the ones which have the uplinks to the core, and then Ian and Toby did the rest. They were then boxed up again to await installation in the decant sites.

Project time estimate for this stage: 2 weeks.

Installation, Testing and Configuration

The switches were installed in the racks and (mostly) patched in by the techs. George and Ian did the initial power-ons, just to verify that things were running as expected. This mostly went as planned, though there was one fibre fault in FH which we called in and which was fixed the next morning by the contractor. The initial framework configurations, which had been set up by George and Ian, were updated by the techs to match their patching, including room and outlet details, based on Gilbert's wiring schedules.

Due to late access, a few temporary patches had to be put in place, for wireless in both sites and for Support/Garry in Forrest Hill. These were later swept up by the techs as they did a final pass through the closets.

Project time estimate for this stage (excluding techs): 1/2 week.

The Move Itself

The move itself was coordinated overall by the "Appleton Tower Decant Working Group". This was initially set up by George, and intended primarily as a technical discussion forum, but was subsequently taken over and expanded by Neil and later Martin to include involved admin staff too. The group met monthly and latterly fortnightly from February to June 2015, with meetings lasting for an hour or so. These meetings were attended by George, Alison, Dave and often Ian.

From April to June there were also building-specific sub-groups, again meeting fortnightly and again for around an hour per meeting (so two hours per fortnight). Again, these meetings were attended by George, Alison, Dave and often Ian. These were primarily intended to coordinate details with the move manager (David Bell)

Taken together there were around 20 meetings of either the full WG or the sub-groups.

The actual move to Wilkie did not take a great deal of CO input, other than what has been mentioned above. For the most part, the office moves were straightforward, and the occupants were expected to do most of the work themselves.

For Forrest Hill, however, the pre-move planning of the standard desks and flip-desks (including lab machines and staff desktops) from levels 3, 4 and 5 was critical to the smooth running of the decant process. There were a number of constraints:

  1. deadline of vacating Appleton Tower by Monday 15th June
  2. start date dependent on on-line exams
  3. Board of Examiners meetings
  4. requirement to keep a suitable number of lab machines available at all times for MSc students.

The timetable was agreed with ISS and the Decant Working Group and was in most cases adhered to as the project progressed.

We encountered a few snags in FH while setting up, including overheating in the Drill Hall, problems with door locks, a substantial leak in the support office (which affected other areas in the building) but still met the deadline of having the machines in the Drill Hall ready for 1st June although this area wasn't opened to students until later that week. The remainder of the Forrest Hill machines were gradually installed over the next few weeks, with them all being essentially done by the end of July. We employed 2 student helpers for a total of 10 weeks which helped to substantially reduce the C(S)O effort required - see separate report for details.

(CO) project time estimate for this stage: 5 weeks.

Bedding In

We of course monitor any new installation quite carefully, to ensure that it is meeting our expectations and to detect and correct any error conditions, and the AT decant was no different. In addition to the AT netmon pages, both George and Ian visited Forrest Hill and Wilkie on numerous occasions to check on progress and to pick up on any teething problems.

Project time estimate for this stage: 1 week.

Total Time

NOTE: We only account here for the CO time spent in the design, planning, setup and installation. There was also a considerable amount of admin-staff, technician and CSO time involved, some of which is noted above. Neither do we account for the productive time lost by the CO/CSO team as a result of their decant, as there are separate projects to cover this.

The overall time for these two projects was 12 weeks, over rather more than a year of elapsed time. Given the complexity of the process, the desire to make the temporary homes as fit for purpose as possible, and the need for business continuity, this actually seems surprisingly reasonable.

We should anticipate the return moves to take a similar amount of effort. There should not be such an initial investment in the design-team meetings required, but there will certainly be liaison with E&B and the M&E contractor to ensure that services (power and data, primarily) are reinstated and fully tested following the inevitable disruption from the removal of the old AT skin and subsequent making-good. Planning for any internal reconfiguration will, of course, be additional to this.

Working documents

 : Units : Infrastructure : Projects : 303-ATFHdecant 

Mini Informatics Logo - Link to Main Informatics Page
Please contact us with any comments or corrections.
Unless explicitly stated otherwise, all material is copyright The University of Edinburgh
Spacing Line