TORONTO – The Canadian government spent $70 million on mainframes last year, despite plans to scrap big iron machines 10 years ago, an executive with Human Resources and Skills Development Canada admitted Wednesday.The mainframes, which in some cases date back to the 1960s, are still using Cobol and other older languages, and the Government of Canada still employs some 600 programmers who keep things up and running. This is on top of the 1,600 applications and more than 650 databases that create massive complexity within the federal public sector, said Andrew Bystrzycki, HRSDC’s director of data management services.
Bystrzycki used these examples as part of a presentation he delivered at the Open Group’s 23rd annual Enterprise Architecture Practitioner’s Conference. Like TD Bank and many of the other organizations which attended the three-day event, Bystrzycki said the Canadian government is struggling to contend with consistent views of citizen data, efficient delivery of services and compliance with privacy legislation. Dealing with technological changes is a big part of the puzzle, he said.
“During Y2K, in the event our insurance systems failed, directors were prepared to stand there and hand out cash,” said Bystrzycki, referring to the so-called Millenium Bug which threatened to bring down older systems on Jan. 1, 2001. “Only the government would come up with a plan like that.”
The government employs more than 20,000 employees and has a budget of around $300 million to manage IT. “But we aren’t agile,” he said, stressing that his opinions were his own and not that of the government itself. “We may say we are, but we’re not.”
Enterprise architecture is seen by many executives as an important blend of IT and business objectives. The Open Group Architecture Framework (TOGAF), now in its 9th version, is designed to help organizations adopt a standard approach to getting its data and systems under control. Mike Lambert, CTO of the U.K.-based Architecting the Enterprise Ltd. And an early contributor to TOGAF, said what began as a prescriptive methodology has evolved into a more flexible blueprint.
“One of the hurdles you see is that people view this as something they just have to get over so you can get on with your job,” he said, warning attendees against adopting a religious-like mindset around enterprise architecture. “Have you ever seen people standing on a street corner taking a verse out of a very big book out of context?” he asked. “That’s not what you want to do.”
The government’s enterprise architecture is intended to ensure Canadians only have to submit their personal information once, so that they could receive the same level of service across all channels, Bystrzycki said. Ideally, Service Canada would not request more information from a citizen unless it involved providing a new service of some kind. Getting to this point means mapping out how the government will deal with structured data in databases, messaging data such as e-mail, documents or other content, and media such as images or video. There has been some progress on adopting a workable approach to metadata, he added, such as shared descriptions for names, addresses and other personal information.
From an IT standpoint, the enterprise architecture involves considerable efforts to reduce duplication, shut down obsolete products and bring down costs, Bystrzycki said. The databases are an example. Bystrzycki said the government runs about 450 Oracle databases, about 115 SQL Server databases and another 50 or more proprietary systems. “Whenever you have that many databases, you know you have a problem,” he said. “They’re massive.”
Even once it has dealt with these areas, however, the government has to address the Personal Information Protection and Electronic Documents Act (PIPEDA) and the federal Privacy Act, Bystrzycki said. For instance, reusing client information means figuring out which version of a citizen’s name and address is the correct one, and how it should be shared and protected. “It’s killing us,” he said.
While data integration might seem like an appropriate goal, Bystrzycki said the government is more focused on data federation -- a single instance of data integration in that the metadata it uses can be employed in the integration processes.