Closure Report
Project Summary
Background
In October 2012, the University of Edinburgh implemented a new eRecruitment service. The service has not been upgraded since the original implementation, and that version (v16) could no longer be adequately supported by the supplier, CoreHR . In addition, this installation had been implemented with a character set that no other eRecruitment customer uses. This put UoE in a unique support position with Core. Core advised that the character set must be converted to the character set the application expects when upgrading. The eRecruitment Proof of Concept project (HRS086) successfully tested converting the character set as part of the upgrade process.
This project set to upgrade the eRecruitment application to the latest version (v21).
Scope
The scope of this project was:
- To convert the character set
- To upgrade eRecruitment to v21
- To update the BI universe following the upgrade
Out of Scope:
- Any enhancements to the service. The upgrade will introduce functional enhancements by default but this project will not be requesting any.
Objectives and Deliverables
|
Description |
Priority | Ouctome | |
|---|---|---|---|
Objective 1 |
To move eRecruitment to the recommended character set |
||
| D1.1 | Converted character set | Mandatory | Achieved |
Objective 2 |
To upgrade eRecruitment to the latest version (v21) |
||
| D2.1 | eRecruitment upgraded to v21 | Mandatory | Achieved |
Objective 3 |
To improve the support position for eRecruitment |
||
| D3.1 | eRecruitment moved on the recommended character set | Mandatory | Achieved |
| D3.2 | eRecruitment upgraded to v21 | Mandatory | Achieved |
Objective 4 |
To refresh and re-engage with the eRecruitment user group |
||
| D4.1 | Representations of all six personas included in the user group | Mandatory | Achieved |
| D4.2 | Updated user manuals and/or guidance | Mandatory | Achieved |
| D4.3 | User training | Mandatory | Achieved |
Objective 5 |
To update the eRecruitment BI universe |
||
| D5.1 | Revised BI Universe if needed | Optional | Achieved |
NB: Delivery of V21 took place on the 26 September 2016 and the upgrade to V21.0.23 was made on 24 January 2017.
Benefits
The main benefit of this project is to provide a stable installation for this key University service which can be enhanced and supported as standard.
| Description | Outcome | |
|---|---|---|
| B1 | eRecruitment using recommended character set | Achieved |
| B2 | Improved support position | Achieved |
| B3 | Maintained and even improved performance against SLA with Supplier | Achieved |
| B4 | New functionality by default in v21 | Achieved |
| B5 | Stable service which can then be enhanced | Achieved |
| B6 | Re-engaged User Group | Achieved |
Success Criteria
The success criteria was set at the start of the project to be that the v21 implementation is a smooth transition for users. This was formally specified by the user group during workshop in April 2016. A set of success criteria for each of the six user personas (Recruiter, Applicant, Originator, Approver, HR Administrator, HR Systems Team) and the ISG support team was defined.
Following delivery, the user group was survey and asked to feedback against these success criteria. Eight replies were received, out of the 24 users on the user group mailing list. Replies for three of the user personas were provided: Approver, HR Administrator and HR Systems Team. The full results from the survey are available at hrs089_survey_results.pdf.
The feedback is overall very positive. We note that one user is “undecided” as to whether the service is reliable and robust. The upgraded service had only been running for a couple of months before the survey, and it is reasonable that the user may need more time to use the system to form an opinion. The only negative point comes from one user who strongly disagree that the training was adequate. This was later confirmed to refer to the documentation provided by Core. This contrasts with the other users who all strongly agree that it was which in these cases refers to the internal documentation.
It is unfortunate that only three user personas answered the survey as all the enhancements brought in by the upgrade were within the personas that didn't answer the survey.
Analysis of Resource Usage
Staff Usage Estimate: 165 days
Staff Usage Actual: 178 days
Staff Usage Variance: 8%
Overall the project closed eight months later than expected. While this has only generated a small amount of work for ISG (8%), the amount of work required from the HR team was significantly more. The upgrading process is complex and Core have a history of missing steps during configuration. The HR team had to run the full set of tests on all environments for every patch and update to the application.
Outcome
Explanation for Variance
The project was delayed after issues were identified soon after the upgrade to v21 was made to LIVE on 26 September 2016. A minor update (v21.0.23) was required in order to fix these issues. It took another four months to have a bug free version of eRecruitment installed, during which additional deployment, testing and management activities were needed.
Key Learning Points
- The use of TestRail and the building of a library of tests helped automate part of the testing, and made it more repeatable for each environment and stage of the upgrade. Any future upgrade of eRecruitment should make use of this and run the full set of tests on all environments DEV/TEST/LIVE. In particular, these should be used during the upgrade of the Oracle Database to version 12c.
- There was a good working relationship between the project team, HR (Susan), IS Apps (Gillian and John), and Core upgrade team (Kelly & Jonathan). The twice weekly calls, shared Jira & HipChat channel worked well at in keeping the communication open and active with Core staff. Although, at time, there were some delays waiting for Core development teams to resolve Jiras, and to build patches and releases.
-
The Project Manager from Nov 2015 to Nov 2016, would like to commend the efforts of the project team, and the patience that everyone showed in working through what was quite a difficult technical upgrade. The team, both UoE and supplier, worked well together and showed great resilience and determination in working to deliver the upgraded service.
- Upgrade should not be scheduled during CAB because it introduces limitations and complications that can be avoided. For example, it was not possible to put up MyEd holding pages during CAB and an alternative measure to lock users out had to be investigated and tested through Core which caused a delay.
- MyEd Launcher channels (especially Java portlets) should be designed so they can easily lock out users, i.e. hide the launcher button and show a holding message. Currently this requires the code to be modified, removed from the load balancers and deployed via Bamboo, and Apache connections restarted.
- Some issues were no identified during testing in DEV/TEST and only became apparent in LIVE. The upgrading process is complex and Core missed different steps during the configuration of the different stages and environments. Because of that, tests that passed in DEV/TEST failed in LIVE. For example, an issue with live templates arose in LIVE and had to be resolved at the last minute. This resulted in Core having to work late hours to complete their tasks to keep us on schedule during upgrade to LIVE. A stricter approach and a better communication of the change control process from Core may help prevent this happening during future upgrade(s).
- Scheduling of an upgrade should allow more than a couple of days between each stage to provide more contingency. Issues and delays meant that the stages ended up running one after the other, with only a couple of hours between two stages in some cases. The team proceeded because it was very difficult to reschedule due to stakeholders availability, and this would have caused further major delays. But this put the team under a lot of pressure.
- In future, all DEV and TEST BI reports should be tested before they are signed off by HR Systems as this was an area that caused unexpected effort in the immediate period after Go LIVE.
- Performance issue with the eRecruitment BI Universe are caused by the database views. This was the case before, and remains after the upgrade. Several of our experienced BI staff and a BI contractor investigated, and the general consensus is that the views are very slow.
- The availability of the project manager was an issue following the secondment of the original PM to Service Excellence early in the summer. This added additional responsibilities to other team members.
- Delays in the delivery of the final fixes, meant a new PM was appointed in November to oversee the last phase of this long project.
- PM or team to notify the team manager of changes to TOIL if differs from that originally approved.
Outstanding Issues
No outstanding issues from this project.