The Cabinet Office is understood to have formally contacted central agencies within the last month and asked them to look for ways to “get rid of Oracle".
No. 10 is believed to be concerned about the amount civil servants are spending on the database giant’s applications and software.
A Cabinet office spokesperson told The Register: "As part of our continuing digital transformation and efficiency programmes, we regularly review technical requirements within a department to see how they may have changed."
The chief problem is the sheer number of Oracle licenses in the UK government, not just their price, although the public sector spent £290m on Oracle in 2013, according to TechMarketView.
Individual IT chiefs will have their own relationship with Oracle and pay for licenses rather than re-use licenses of those within their own department.
In January The Reg reported that the Department for Environment, Food and Rural Affairs (DEFRA), which has around 10,000 staff, was forking out for two million Oracle licenses at £155 per employee, for an annual cost of £1.3m per year. (That worked out at 200 licences per civil servant in the department.)
That’s contrary to the Cabinet Office’s own guidelines of £93 on licenses, with a view to reducing that down further to £52.
Meanwhile, the mighty Home Office has tried to slash its Oracle budget by moving an ERP contract to a shared-services platform run by Steria. A Register source told us: “Nobody has a holistic view or how Oracle is used across the whole government or looking at economies of scale.”
The source described the the central communication as an “edict” that has been interpreted as an order to move away from Oracle.
(Score: 2) by NCommander on Thursday August 06 2015, @08:51PM
A fair bit of the problem is that Oracle's has a "unique" dialect of SQL with different types compared to other databases. Furthermore, most Oracle projects I've seen (ab)use plSQL quite heavily. There's an entire market of things that put plSQL on other databases, such as EnterpriseDB which creates pgplSQL. While I have no recent Oracle experience, I don't remember there being a whole lot of "free" (aka, part of the database) tools for locating and doing query optimizations which added a lot of pain.
As such, something that was written for Oracle tends to be a nightmare to port anywhere else. In addition, Oracle has a LOT of overhead and pain to not only get running, but get running well. It may be better now, but Oracle 10 was a massive nightmare just to get the database installed, let alone usable on localhost to test some software. I've heard horror stories about RAC and other add-ons to the database; the idea of a clustered Oracle gives me chills (mySQL cluster is bad enough ...)
Still always moving