The cost of failing an Oracle ULA certification: a $4.5M lesson
A well-run bank did everything it thought it needed to — and still failed its Oracle ULA certification, then renewed for US $4.5 million to cover licences it didn't need. This is the story, the mistakes, and how the same estate was later certified cleanly.
The cost of failing an Oracle ULA certification can be very high. This is the story of a well-known bank — not a very large one — headquartered in a major Indian city, and how a certification everyone expected to be routine went badly wrong. The timeline is simple: ULA at Year 0; a decision to certify at Year 3; a failed certification; a renewal at Year 4; a fresh Oracle compliance approach at Year 5, when they called us; and a clean certification, finally, at Year 6.
01A ULA bought to be certified
At Year 0 everyone was happy. The Oracle sales team were happy. The CIO was happy, because he could now give the applications and infrastructure teams complete freedom to deploy any number of Oracle database licences. The DBAs were happy, because they no longer had to wait for a fresh purchase before installing a new database. Those were the best of times.
There was one crucial detail. When the bank bought the ULA, it had already decided it would exit the agreement at the end of the three-year term and certify. That intention was the very basis on which the bank's executive committee approved the expenditure in the first place.
02The certification that failed
Fast forward three years, to one month before the term ended. The IT team asked Oracle for the certification process. The sales team were not pleased and pressed for a renewal instead — there was revenue, and commission, in a renewal. But the bank was clear: it would exit and certify. The account was handed to Oracle's LMS/GLAS team to run the process.
The bank was given a questionnaire and the Oracle Server Worksheet, a spreadsheet capturing the entire server, datacentre and virtualization architecture along with the Oracle installation details. The bank filled it in happily — after all, it knew exactly what had been installed, when, and why. It calculated its licence counts and felt satisfied it had made good use of the investment. The numbers weren't enormous, but they were enough to justify the original spend.
03What Oracle handed back
After a few weeks reviewing the declaration, the LMS/GLAS team asked for a meeting. They wanted to run scripts on a few selected servers, asked whether every Oracle deployment had been declared, and asked whether there were any cloud deployments. The bank agreed to the scripts, confirmed everything had been provided, and confirmed there were no cloud deployments. Then the findings came back, and the bank was in shock:
| PRODUCT SET | WHAT ORACLE PROPOSED TO GRANT |
|---|---|
| 3 ULA PRODUCTS | A potential grant of only up to 20% of the expected licence numbers. |
| 2 ULA PRODUCTS | A potential grant of zero. |
| 2 PRODUCTS NOT IN ULA | A demand for additional licences across every core in the virtualized environment — about 10× the proposed renewal price. |
This is the moment the bank realised its certification was failing badly — and began, too late, to understand why an expert should have been involved from the start.
04The $4.5M ‘safer’ renewal
At the start of Year 4 the bank began talking to the Oracle account manager, who was suddenly very friendly and keen to help the bank avoid spending so much money. He even offered a discount on the renewal proposal. Weighing a discounted renewal against a long and uncertain battle, the bank judged it safer to renew. The renewal cost US $4.5 million over three years. The estimated additional Oracle usage the bank actually needed was zero.
05How it was finally fixed
The bank called us at Year 5, when Oracle approached again about a separate compliance matter. Reviewing the earlier renewal, we were not surprised to find that Oracle's draft report had contained several errors of interpretation — and that the bank had also made mistakes in its own declaration. We explained both. The bank saw what it could have done differently, but that time had passed.
So we focused on the next term. As the second ULA came to an end we helped the bank optimise the deployment, maximise the potential licence grant, remediate any non-compliance, and drive through a clean certification. The outcome: the bank is now on track not to spend a single extra dollar on fresh Oracle database licences for at least the next five or six years. The whole $4.5 million detour had one cause — a certification treated as a form to file at the end, rather than a programme to run from the beginning.
06Straight answers to the questions we get most
FAILING A ULA CERTIFICATION — PLAIN-LANGUAGE FAQ
What does it mean to “fail” a ULA certification? Oracle grants far fewer perpetual licences than you deployed, and often demands new licences for products or usage it argues were outside the ULA's scope. Faced with the gap, many customers feel forced to renew — exactly the outcome the process is built to produce.
How much can a failed certification cost? In this engagement, a bank renewed for US $4.5 million over three years despite needing zero additional Oracle usage. Done properly, that spend was entirely avoidable.
Why did Oracle ask for licences on products not in the ULA? Because usage had drifted outside the ULA's defined product set, and Oracle counted the exposure across the entire virtualized core estate — producing a demand many times the renewal price.
Can a failed certification be recovered later? Yes, but you lose the years in between. In the next term we optimised, maximised and remediated the estate and secured a clean certification — putting the bank on track to spend nothing further on Oracle database licences for years.