Completion Report
Project Summary:
The stated objectives for this project were:
No. | Description | Objective Met? |
O1 | Upgrade wiki service to the latest version of Confluence | Y |
O2 | To fully test the upgraded service and ensure it is fit for purpose | Y |
| O3 | Investigate how Grouper and mobile issues might be addressed with the newer version of Confluence | Y |
In the project brief, the following deliverables were identified:
No. | Description | Delivered? |
D1.1 | Upgrade wiki DEV, TEST and LIVE environments to version 5.7 | Y |
D1.2 | Replacement servers to host the DEV, TEST and LIVE environments | Y |
| D1.3 | Updated technical and support documentation | Y |
D2.1 | Load testing of new service | Y |
| D2.2 | DR testing of new service | Y |
| D2.3 | User acceptance testing of upgraded service | Y |
| D3.1 | Reports on Grouper and mobile issues, with recommendations for next steps as part of analysis phase | Y |
All of the objectives and deliverables have been achieved, though under deliverable 2.1 (DR Testing) it should be noted that a full test was not carried out as we have now moved the DR capability to a technology and procedure (Veeam) that has been tested before. The project has therefore delivered a new DR plan for this service.
Analysis of Resource Usage:
Staff Usage Estimate: 41 days
Staff Usage Actual: 68 days
Staff Usage Variance: 66%
Other Resource Estimate: 0 days
Other Resource Actual: 0 days
Other Resource Variance: 0%
Explanation for variance:
Extended duration
This project has encountered several delays, with the original suggestion for deployment into Live of late May being moved out and eventually achieved in the first week of August. However, for reasons that will be explained below, the remaining sign-offs were also further delayed.
The first delay to the original milestones was caused by the Build stage overextending (issue 2) and having to wait until colleagues were available to continue work on it. This was attributable to the scheduled work taking longer than estimated. The next recorded delay was issue 3 - the final pinning down of a date that was suitable for the deployment of the upgrade into Live. It had always been known that the original suggested date for this was provisional, and that it would be inevitable that a delay would be required when it came to setting a date that suited all participants and requirements. The date for going live with the upgrade was set, at this point, as 2nd July.
The next delay was not anticipated and was a result of the upgrade not achieving acceptable pass results in the load testing carried out at Acceptance. This problem was eventually resolved through work done by Alain Forrester and Riky Harris, in conjunction with Atlassian, on the 'Sections' functionality. Once this was sorted out (issue 5), a new date of early August was planned for the changes in Live. This date was met and the necessary migrations and upgrade were completed through out of hours working by the relevant members of the project team. It was after this that the remaining milestones were delayed by the problems encountered after deploying the changes into LIve. The reported issues led to the formation of a monitoring group by the Head of Production Management that included colleagues from Development Technology, Service Management, Applications Management and Technology Management. This group regulated the wiki performance, stepping in when necessary to resolve any issues, over the remainder of August, with an acceptable level of stability gradually being achieved.
Additional effort
| Stage/Task | Estimate (days) | Actual | Difference |
|---|---|---|---|
| Project Management | 7.0 | 13.8 | 6.8 |
| QA/Meetings | 1.5 | 3.0 | 1.5 |
| Planning | 2.0 | 2.3 | 0.3 |
| Systems Analysis & Design | 5.9 | 8.1 | 2.2 |
| Build | 8.9 | 4.3 | (4.6) |
| Integration | 8.8 | 11.3 | 2.5 |
| Acceptance | 3.2 | 5.1 | 1.9 |
| Deployment | 3.1 | 20.6 | 17.5 |
| Closure | 0.6 | 0.8 | 0.2 |
| Unplanned Activity | - | 1.5 | 1.5 |
| Total | 41.0 | 70.8 | 29.8 |
The original estimate for the project was 41 days, but the final figure is 70.8 days overall. A PICCL was raised in August (27/08) to cover the additional effort that had been used to date, and anticipate the final work of the recovery group; at this point, the final figure was expected to be 68.5, and this was approved by all concerned. However, some late timesheets were included and an additional 2.3 days (covering monitoring & service restarts in August/early September) has subsequently been recorded.
The main differences between estimates and final, actual figures can be explained as follows:
- Project Management - this has risen as a natural consequence of the extended duration and added complexity of the project.
- Unplanned/QA - this has increased because of similar reasons to those listed above
- Integration - attributable to the need to rerun load testing
- Acceptance - again, this has risen because of the need to undertake testing a second time.
- Deployment - the extra effort for this stage includes both the time required to work on the migrations and upgrade required for the LIVE environment, which needed additional out of hours work, and the various tasks undertaken by the recovery team:
- Additional work on deploying upgrade to Live - 5d
- Work by recovery team w/c 10/08 - 4d
- Work by recovery team w/c 17/08 - 5d
- Work by recovery team w/c 24/08 - 1.4d
- Work by recovery team w/c 31/08 - 2d
As noted in the PICCL linked above, it had been agreed across the Senior Management Team that the effort to cover the recovery team work could be included in the total for this project.
Key Learning Points:
A number of points have been raised since the deployment of the upgrade; these have covered:
1. The next upgrade project should encompass work on Grouper/connector aspects of the service more fully. This project carried out some analysis, but did not add these to its scope.
2. We should upgrade our relationship with Atlassian in order to help escalate any further issues with upgrades.
Outstanding issues:
At the time of preparing this report, some issues were still open. These are:
1. The functionality to move pages: this has been disabled (by being hidden) until a fix has been received from Atlassian. Moves can be done in small parts via the Wiki service.
2. Some of the wiki plug-ins have been disabled, and continue to be reviewed.
Please see the report on the recovery group for more detail.