A$100m supercomputer project queried
Auditor critical of Uni of Melbourne cluster of problems
A A$100 million supercomputer facility in Victoria, funded by A$50m of government funds with the rest coming from the University of Melbourne, has come under fire from that state’s Auditor-General.
The VLSCI – Victorian Life Sciences Computation Initiative – is due to be completed in 2013, and would create one of the world’s top-ten facilities. Stage 1 of the project, a smaller IBM cluster, went into operation in August of 2010. The final facility is intended to have clusters from both IBM and SGI.
However, the Auditor-General’s report casts doubt on whether the facility would actually meet researchers’ needs.
The VLSCI project was conceived by the University of Melbourne and successfully pitched to the government. However, the auditor says “the absence of sufficient needs and options analysis and sound procurement processes” means that the University can’t show if the money will be well spent.
According to the report, thee university approached the market before it had found out what researchers actually needed, and it can’t demonstrate value for money.
The 400T flop facility was announced amid much fanfare in 2008. However, the change of government since then led to the referral of the project to the Auditor-General.
The risks of allowing haste to overcome prudence are highlighted in the report, which says that although the proposals received in March 2009 did not meet all of its requirements, the University preferred to work to reduced scope rather than repeat its approach to market. The proposals received in 2009 were also over budget by more than A$6m.
The report also suggests that IBM’s close relationship with the university gave it advantages in the tender process. While not naming IBM, the auditor says “the company selected as the university’s partner was involved in the project since 2007.”
This gave it “an intimate knowledge of the university’s requirements and input into the design of the initiative; time to determine how it could assist the university deliver the initiative; and extensive access to university staff.”
Without suggesting impropriety, the report notes that there was “little recognition of this advantage in university documents”.
IBM’s success at relationship-building in Victoria has also been demonstrated by its decision to locate a 150-staff R&D lab in Melbourne.
The report also notes that the contracts were structured in such a way as to give the university little scope to change vendors between its first-stage rollout and the larger second stage.
Although the university believed it could either vary its requirements or choose a different vendor between the two stages, the report concludes that the contract “provides for the automatic expiry” of its agreement “on the termination of the PCF fit out agreement … As a result, if the University wants to continue its research partnership, it does not have the option to engage a different company to deliver stage 2”. ®
Sponsored: Hyper-scale data management