Open-source software helped London's Oyster card system move past a proprietary roadblock, an open-source conference in London was told last week.
The Oyster contactless card system, which handles payments for travel on London's buses and Tube system, suffered from lock-in to proprietary systems, which hindered developments to the online payment systems, said Michael Robinson, a senior consultant with Deloitte, at the Open Source Forum event in London. "The hosting was on a proprietary system, centred on one application," he said. "It demanded certain hardware, and was locked into one design of infrastructure."
Ninety percent of all bus and underground travel in London is paid for by the Oyster RFID (radio frequency identification) cards, which began trials in 2002, and were launched by Transport for London (Tfl) in 2003. With Oyster fares generally cheaper the comparable cash fare, it's no surprise there are now 12 million cards in circulation.
Despite this success, Tfl wants more people to move to online payments and automatic top-ups, which would reduce the demands on staff and machines at stations in ticket offices. Currently, 1.4 million Oyster cards are held by registered online customers, but Tfl wants to increase this, Robinson said. "Smartcards have even more benefits when topped up online — there is no queue, and it is self service."
But expanding the online system has been a problem. Early in its life there were ambitious plans to add cashless payments to the Oyster card, making it an electronic wallet, but the idea was shelved in 2005.
Read this
Feature: Ten things you can do to help open source
Ask not what open source can do for you and your business, but what you can do for open source...
Three-and-a-half years into the Oyster system's life, it became clear to Tfl there was an upgrade that could not be put off, said Robinson. Simply put, the online system wasn't up to the job. It was too expensive, did not give good enough quality of service, and was not responsive enough to support the kind of promotions Tfl wanted to use, he said: "It couldn't respond to business changes, and it didn't scale."
Oyster also faced another problem: the site had to be upgraded to meet the payment card industry's PCI DSS requirements for security, which had emerged since the scheme began.
The Oyster system, including the scanners in buses and underground stations and the back-end database, is run for Tfl by services company EDS, but Tfl turned...
Talkback
How can switching to open source save them any money? Haven't they read any of Micro$oft's studies showing that windoze is the fastest, most innovative, and lowest TCO, in the industry? I'm sure Redmond will have a different version of this story.
"It could have been done with MySQL or Postgres, but that would have taken more consultancy time"
The question is whether hiring a competent MySQL / PostgreSQL consultancy to do the work would have cost more than Oracle's licensing fees.
It would be interesting to know the nature of the Deliotte / Oracle UK relationship. Audits, anyone?
That is a good point, and one I'll certainly follow up if I get to return to this story.
The impression given was that Oracle included pre-built features, that would have taken time and money to develop and support in other databases.
Whether that is true, and whether the time and money would actually have been greater than the Oracle fees, is part of the usual grey areas in these sort of projects.
I suspect it was more to do with TfL and Deloitte's comfort zones - and also with Deloitte's Oracle relationship - and also the necessity for Deloitte to put together a scheme that makes money for them, and which TfL is prepared to buy into.
Peter
I have now had an email from someone at Deloitte, sounding mildly irritated at our guesswork around Oracle fees.
"The implied suggestion is that TfL had to incur a cost to obtain an Oracle license or support," we are told. "That's not necessarily true - it's quite feasible for an organisation of that size to have an 'unlimited' type of licensing arrangement for one vendor's products."
Perhaps everyone else knows this, but that sort of licensing is actually a new thing to me. It would indeed explain TfL's use of Oracle much more simply than my guesswork. It's there, it's got the features, and it's paid for, so they used it.
Unlimited licences certainly do exist. Here's Gartner advising on them.
http://www.gartner.com/DisplayDocument?ref=g_search&id=498258
I'm now interested to know how they work, both in practical terms and in terms of how they shape the market.
If memory serves the Deliotte consultant claimed to have undertaken an assessment of the alternatives and chose Oracle because it had more relevant features.
Perhaps I can make two points:
Firstly, 'unlimited licensing' arrangements with Oracle must be reassuringly expensive unlike PostgreSQL or MySQL that offers 'unlimited licensing' for nothing.
And secondly, if TfL had unlimited access to Oracle why did Deliotte even bother to test the suitability of PostgreSQL and MySQL as claimed?
Thanks - both those questions occurred to me.
However, TfL and Deloitte are free to use whatever technology they want to use, whatever they pay for it, I would have thought.
And open source people are free to point out what they believe would be a cheaper option.
Maybe Deloitte bothered to evaluate open source databases (your point 2) precisely because they were free (your point 1).