PhaseAReport

(9) Financial and Management Report

(9.1) Introduction

This section of the Phase A Report outlines the way we have chosen to manage the AstroGrid project, and summarises expenditure on personnel and finances up to the end of Aug 2002.

Other parts of this report (Architecture Overview and Phase B Plan) have described the Unified Process (UP); in this report we will outline its impact on the management of the project and the implications for a work package approach to project structure. AstroGrid non-salary finances have been managed centrally as opposed to the usual up-front distribution to institutes and we explain how this was set up and controlled.

In the rest of this report, we refer to the four quarters of Phase A (plus a fifth quarter extension to Phase A which we have requested of PPARC). The actual dates of these quarters are:

• Q1: Sep'01 to Nov'01
• Q2: Dec'01 to Feb'02
• Q3: Mar'02 to May'02
• Q4: Jun'02 to Aug'02
• Q5: Sep'02 to Dec'02 (yes, four months long)

Note that expenditure in Q1 included, by PPARC's permission, travel expenditure undertaken in the project planning phase before the official start of Phase A. Also note that some aspects of Q4 expenditure are not yet known, as invoices from institutes have not yet been received.

(9.2) Management

(9.2.1) Project Policy and Oversight

Overall responsibility for the project rests with the AstroGrid Lead Investigators (AGLI). In alphabetical order, they are :

• Peter Allan (RAL)
• Simon Garrington (Jodrell Bank)
• Louise Harra (MSSL)
• Andy Lawrence (Edinburgh, Project Leader)
• Richard McMahon (Cambridge)
• Fionn Murtagh (Belfast)
• Mike Watson (Leicester)

The AGLI set policy and direction, monitor the progress of the project and the performance of the Project Manager and Project Scientist, and make decisions on resource allocation. Monthly telecons are held which also normally include the Project Manager and Project Scientist. One member of the AGLI is elected as Project Leader, who acts with the Project Manager and Project Scientist as a small executive team. Nearly all day-to-day management is delegated to the Project Manager. The Project Leader takes an overall leadership role, and is responsible for liaison with PPARC. AstroGrid is overseen by two bodies - PPARC's Grid Steering Committee, which holds the budget which funds AstroGrid, and the AstroGrid Oversight Committee.

(9.2.2) Work Structure

In AstroGrid Phase A we have had both a traditional workpackage structure, and a modern flexible incremental approach using the Unified Process. Here we explain how this came about and how it works.

The majority of medium to large sized academic projects, especially those distributed across several institutes, tend to have a work package structure. A work package is focussed on one aspect of the project - an area of technology or research - and led by an expert in that subject. The work packages cooperate on standards and work together to ensure that their products integrate correctly. This was the way that AstroGrid was conceived and set up, with the following Phase A workpackages (8):

• WPA0 Project Management and Infrastructure
• WPA1 Science Requirements Analysis
• WPA2 Data Grid Technology
• WPA3 Storage/Compute Technology
• WPA4 Database Technology
• WPA5 Pilot Programme
• WPA6 WFCAM/VISTA Liaison
• WPA7 External Liaison
• WPA8 Phase B plan

The workpackages were used for formal project control (see below) but the actual work carried out differed in some respects. Most importantly, a large amount of work has formally been in Architecture development (see below), which was formally reported through A0. Package A6 (WFCAM/VISTA development) soon ceased to be a formal AstroGrid workpackage, although such liaison remains important. While our effort deployment was flexible, much of the work completed by the project did reflect the original structure, as can be seen by the fact that chapters of the Phase A report have similar titles to the above.

In recent years, medium to large software projects in the commercial world have adopted different methodologies (now being referred to as Agile methods) which emphasised an iterative, incremental approach to system building. These methods drastically increase a project's chance of being successful.

AstroGrid hired, as its project manager (PM), Tony Linde, with long-term experience in commercial applications development. After he had explained the Unified Software Development Process (UP) (1), one of the new methodologies, the project team decided to adopt the UP as its project development methodology.

As different as these two approaches seem, there was little long-term conflict during Phase A. This is because the initial stages of the UP are concerned with the development of an Architecture (see Architecture Overview document) which would represent a high level model of the proposed system. The responsibility for development of the architecture devolved to the PM and was subsumed into work package WP-A0 (Project Management and Infrastructure).

The only area of conflict came in the early days of development of the architecture when team members within other work packages were needed, at some length, to work on the architecture. This was resolved by putting together a small team to concentrate on the architecture with other team members' involvement confined to reviewing the models and other documents produced and taking part in periodic focus meetings where direction was discussed and set.

(9.2.3) Communications

This also touches upon another important aspect of the management approach: project communication. From the beginning, the project had a web site with several pages relating to background and project structure as well as an internal mailing list. The new PM, with the agreement of the AGLI and project members, implemented several additional measures which opened the majority of project communications to anyone who chose to view them, though the mailing list was still occasionally used for private, group-wide communications. This was somewhat akin to the software industry's Open Source initiative, which we therefore nicknamed Open Science (2).

The three initiatives (in addition to a revamped website at http://www.astrogrid.org/) were (3):

• News:
  The News site (http://news.astrogrid.org/) is most often used for announcements to the whole project team (replacing the main use of the mailing list). Updates can be posted several times a day and content includes news stories, meeting, conference and other event announcements and polls. Any registered user can create these items and most allow the opportunity for feedback to be added.
• Forum:
  The Forum (http://forum.astrogrid.org/) provides an area for topic-based discussions to develop as well as a place for novices to ask questions and those with problems to seek solutions. Again, any user can create a new topic or can add comments to existing topics.
• Wiki:
  The Wiki (http://wiki.astrogrid.org/) is the most unusual and exciting area of the site. A wiki is a form of website that allows the registered user to change the content of the site, adding comments or new sections to existing pages or creating their own pages. The site includes help pages, a tutorial and an experimental wiki where users can try things out without fear of damaging existing content. It is divided into a number of webs: one for the AstroGrid project, general webs for VO, Grid and e-science topics, and support webs such as the tutorial and test pages.

These online facilities helped with keeping team members in touch with developments in other areas of the project but, with such a widely distributed project, it was equally important to ensure people had opportunities for direct contact.

The original plan was to have full consortium meetings, bringing together everyone involved with the project, every three months. This proved unmanageable, so it was decided that the consortium meetings would be held every six months: Sep'01, Dec'01, Jun'02 so far and the next will be in Jan'03.

At six month intervals, the consortium meetings were unsuitable as a means of keeping people up to date with progress with ongoing work. We therefore instituted Focus Meetings (4). These were subject specific meetings in which a small group of people interested in the subject could work together. The focus meetings were highly successful in developing the science problems, use cases and architecture.

(9.2.4) Control

Control of project activities was managed through the Work Package Managers (WPMs) each of whom reported on a quarterly basis (5). A meeting of all WPMs would be held just before the end of a quarter to review progress and plan tasks for the coming quarter: during the intensive architectural activity, this meeting would be subsumed by an architecture progress meeting, but the goals and ends were the same.

A quarterly Forecast Report contained:

• Summary:
  A textual summary of the goals set for the coming quarter.
• Resources:
  People and equipment which would be deployed on the work package activities.
• Planned Tasks:
  A coded list of tasks (codes were retained from previous quarters if they overran) showing planned start and end dates and the people who would work on the task.
• Deliverables:
  List of concrete deliverables that would come out of the tasks with due dates.

The quarterly Progress Report on the previous quarter's activities contained:

• Summary:
  A textual summary of the achievements and issues arising from the past quarter.
• Resources:
  People and equipment which was actually deployed.
• Tasks:
  A list of the tasks from the forecast plus any unplanned activity. Each task showed actual start and end dates along with percentage complete.
• Deliverables:
  List of actual deliverables produced with links to where they might be found.

As with everything else on the AstroGrid project, these reports were freely available on our web sites. Only personnel-related information and other sensitive material was kept offline.

(9.3) Personnel

In addition to the Lead Investigators, the AstroGrid team contained the following project funded personnel at the Sept 2001 start of Phase A :

Location	Name	Position	FTE Aug 2002	sm to Aug 2002
Leicester	Clive Page	RA/Developer	1.0	10
Cambridge	Guy Rixon	RA/Developer	1.0	12
Edinburgh	Bob Mann	RA	0.5	6
MSSL	Bob Bentley	RA/Developer	1.0	12
Edinburgh	Clive Davenhall	RA/Developer	0.5	6
RAL	David Pike	RA	0.5	6
RAL	Chris Perry	RA	0.5	6
Jodrell Bank	Anita Richards (to end March)	RA	0.0	3
Jodrell Bank	Ant Holloway (from 1-April)	RA	0.2	1
RAL	John Sherman	Co-ordination	0.2	2.4
RAL	Dave Giaretta	RA/Developer	0.2	2.4
RAL	Peter Allan	Lead Investigator	0.1	1.2
TOTAL			5.7	68

Note-1 : Ant Holloway replaced Anita Richards at Jodrell when Anita took up the AVO post below.

_Note-2 : FTE=Full Time Equivalent is the fractional effort as at August 2002. This is a rate of expenditure, so that someone rated at 0.5FTE is expending 50% of their effort on AstroGrid.

Note-3 : sm=staff months is the integrated effort expended to the end of August 2002. Likewise sy=integrated staff years, so that someone working at 0.2FTE for 3 years has expended 0.6sy.

The Lead Investigators, as academic staff, are not funded by the project, except for Peter Allan. (The RAL system requires charging for all effort used). All these staff were funded by PPARC. Funding for the above staff was routed through a variety of existing grants, except for RAL staff, who are funded through the PPARC-CLRC Service Level Agreement.

New recruits to the project were:

Location	Funding	Name	Position	Start Date	FTE Aug 2002	sm to Aug 2002
Cambridge	PPARC	Nic Walton	Project Scientist	November 1	1.0	10
Leicester	PPARC	Tony Linde	Project Manager	November 1	1.0	10
Jodrell Bank	AVO	Anita Richards	RA	April 1	1.0	5
Leicester	PPARC	Tim Goodwin	Web Developer	July 1	0.5	1
Leicester	PPARC	Patricio Ortiz	RA	July 1	1.0	2
Cambridge	PPARC	Kona Andrews	Developer	July 8	1.0	2
Edinburgh	AVO	Martin Hill	Developer	Aug 1	1.0	1
Edinburgh	AVO	Alan Maxwell	Developer	Sept 2	1.0	0
TOTAL					7.5	31

Recruitment of the later RAs was seriously delayed by issues relating to the funding of AVO posts, which had a knock-on effect of delaying the recuitment of the PPARC posts. This was finally resolved by switching one of the AVO posts from Cambridge to Edinburgh, and one of the PPARC posts in the reverse to provide the above. All posts on the project to date are of three year duration.

The size of the team (excluding the AGLI) is now 18 people, adding to 13.1 FTEs. Given the start dates, the effort expended to the end of August 2002 (including the cost of Peter Allan) is 8.25sy.

(9.4) Finances

The total Phase A expenditure up to the end of August 2002 has been approximately £661K. (This is the PPARC expenditure only).

AstroGrid finances have been split into personnel costs and non-personnel costs, so that the non-personnel costs could be centralised into a single central budget controlled by the Project Manager in Leicester. This was done in order to keep flexibility in non-personnel costs, and to make sure the PM had adequate control over expenditure.

(9.4.1) Personnel Costs

Staff costs have been expended through a variety of grants to different institutes. For University based staff, on top of salary, these costs include standard fractions of secretarial and system management support staff, and the standard PPARC grant overhead of 46%. Some of the grants concerned had other minor costs attached. For RAL based staff, time is charged at a uniform staff rate agreed by negotiation between PPARC and CLRC. A detailed breakdown has been provided to PPARC.

The total funded staff effort over the year has been 8.25sy, some of which, the AVO funded staff effort, is at zero cost to PPARC. Grant claims are not all made yet so the accurate final cost is not known, but our out-turn forecast for the one year staff-related costs to PPARC, including all the above related costs and overheads, is £492K.

(9.4.2) Non-personnel Costs : Financial Control Mechanisms

Non-personnel expenses were provided by PPARC through three grants, two for capital equipment and one central budget grant. These were seen however as representing a single budget controlled by the PM, who set out an overall budget plan in November 2001. Procedures were then put in place for managing these finances.

The procedures made each WPM responsible for forecasting expenditure during the coming quarter. These forecasts were reconciled and adjusted (if necessary) by the PM and then combined into the budget for the next quarter. Any team member needing to travel or make a purchase on behalf of the project claimed for the expenditure from his or her institution.

At the end of a quarter, each institution collated all claims made and invoiced Leicester for the total amount. The PM checked these invoices against the budget and authorised payment. It took some time to get this process established in the consortium institutes but it now seems to be working well. Invoices up to and including Q3 have been received from all institutes and two, to date, have been received for Q4.

Finances, as they are known at the time, are submitted to meetings of the AstroGrid Oversight Committee (6) for approval.

(9.4.4) Non-personnel costs : Expenditure Report

The current forecasted expenditure (based on invoices submitted to date plus forecasted expenditure for Q4) is:

Of these items, two stand out:

• Travel:
  This figure is difficult to predict. Invoices to date only total £55K but Q3 and Q4 involved large amounts of travel by many of the team members. The reason for this is two-fold. Firstly, it is the traditional season of conferences in the academic community. Secondly, AstroGrid is now seen as a major player in the VO and Grid domains. We therefore were invited to attend and speak at many more meetings and conferences and, in order to maintain our lead in certain areas (authorisation, data access and mining, astronomical ontology, to name a few) and to keep up in other areas, we sent team members to other meetings.
  We anticipate that the final actual expenditure on travel will be much the same as that budgeted.
• Equipment:
  More was spent on the Leicester Beowulf cluster and Cambridge Sun server than was budgeted. This was due to expenditure being less in other areas than expected. We decided to use the underspend to increase the number of nodes in the cluster from 4 to 8, enabling a more realistic investigation of cluster-based data mining techniques, and to purchase a newer type of Sun server. Both of these machines were also upgraded to have around one terabyte of disk storage to enable the loading of complete data archives.

Phase A has been extended into a non-standard, four month Q5. This will allow the team time to complete the architecture and the demonstration projects. The budget for this extended quarter will be £15K for travel but may be further extended after the meeting of the Grid Steering Committee to consider the proposal for Phase B funding.

(9.6) References

(1) Unified Process: See the more detailed description in the Architecture Overview, elsewhere in this Phase A report.

(2) See also http://wiki.astrogrid.org/bin/view/Escience/OpenScience.

(3) Also described in PPARC Frontiers magazine, Issue 13, p26-27

(4) For a schedule of focus meetings, together with reports from them, see http://wiki.astrogrid.org/bin/view/Astrogrid/FocusMeetingNotes.

(5) All forecasts and reports are linked from the wiki page: http://wiki.astrogrid.org/bin/view/Astrogrid/WpReports

(6) AGOC: committee set up by PPARC to oversee the progress of the AstroGrid project. This committee has met twice so far; for reports from those meetings, see: http://wiki.astrogrid.org/bin/view/Astrogrid/OversightCommittee.

(7) The advertisement for this position (for which applications closed on 13th September) can be found at http://www.le.ac.uk/personnel/jobs/p9048a.html and a full description at http://www.le.ac.uk/personnel/jobs/p9048.html.

(8) Detailed descriptions of the work packages can be found by following links at http://wiki.astrogrid.org/bin/view/Astrogrid/WorkPackages.

-- TonyLinde - 02 Oct 2002 -- AndyLawrence - 01 Oct 2002