IRS Data Centers Gird for Tax Season

irs-presentation

Tony McMahon, who oversees the data center infrastructure for the Internal Revenue Service, discusses the agency's operations earlier this month at Data Center World in Nashville while colleague Linda Riess listens.

It's crunch time for the Internal Revenue Service. As the IRS ramps up for the annual crescendo of returns around the April 15 deadline, the state of the agency's infrastructure depends upon who you ask. 

IT executives at the IRS say it has made huge strides in modernizing its data centers,  which processed 139 million returns and issued $298 billion in refunds in 2009. Independent tests say the IRS web site is the fastest U.S. government site, and one of the fastest on the web.

But a key government watchdog, the General Accounting Office, says the modernization effort hasn't moved quickly enough, and continues to fault the IRS for weaknesses in its security practices. 

Electronic Filing Key to Efficiency
What's not in dispute is that the agency's growing capacity to process electronic returns is critical to improving the efficiency and accuracy of the nation's tax system.  

In a presentation earlier this month at Data Center World, Tony McMahon of the IRS said there are widespread misperceptions about the agency's IT operations.

"One of the things you often hear is that IRS is running on 1960s technology," said McMahon, the director the Enterprise Computing Centers for the agency. "We are not. We actually have state of the art technology in our environment." That includes IBM z10, z800 and z900 mainframes, as well as Unisys 6800 systems.

Major Data Center Consolidation
The IRS has recently retired seven data centers and a ton of legacy equipment through a data center consolidation. The agency, which once had 10 data centers, now operates major data centers in Memphis and Detroit to support the agency's primary site in Martinsburg, West Virginia.

The Detroit site supports administrative computing, while most data is processed in Martinsburg and backed up to Memphis. These facilities process 7 million transactions per day, and run more than 4,500 batch jobs daily as they exchange data with other agencies.

That processing capability comes into play when the government has a pile of checks to distribute quickly. The IRS infrastructure was used to distribute checks for the Obama administration's stimulus program, and provided emergency check processing capacity when the "Cash for Clunkers" program encountered lengthy delays last year, McMahon said.

Split Operations
The agency's operations remain split between a legacy system and a newer design known as the Customer Account Data Engine (CADE). Last year CADE processed 40 million returns, up from just 7 million in 2006. But that was still behind schedule, leading the IRS to put CADE on hold after investing five years and more than $400 million in development. The agency is now focusing on a successor system known as CADE 2.

McMahon said CADE 2 will be able to process electronic returns and issue refunds in 1 to 2 days, down from 7 to 8 days for the original CADE system and six to eight weeks for the systems that process paper returns. McMahon said the new system is the top priority for the agency's new Chief Technology Officer, Terry Milholland, who previously worked for VISA.

Capacity and speed weren't the only drivers behind the IRS decision to rethink CADE. IRS officials say the need to enhance security and incorporate new technologies also influenced the decision to rethink the CADE strategy.

Tradeoffs in CADE Overhaul
"Stopping CADE development has trade-offs in that IRS will not be able to materially increase the number of returns processed on CADE during the 2010 filing season, which, in turn, means that the number of taxpayers benefiting from faster refund processing will not increase," the GAO noted in its review of the 2009 filing season. "On the other hand, IRS’s new strategy for modernizing individual taxpayer accounts is intended to address the risks and challenges of the initial approach."

The GAO has been a consistent critic of IT security practices at the IRS, saying the agency "needs to continue to address significant weaknesses" in a new report this month. The GAO said the IRS continued to make progress addressing its security issues, many of which involved the use of clear-text passwords and mainframe configuration.

But the GAO also cited ongoing problems with updating security patches and policies regarding user password strength, which the GAO said "continue to jeopardize the confidentiality, integrity, and availability of financial and sensitive taxpayer information."

On the security front, McMahon says the agency has significantly beefed up its focus on intrusion detection. "We've never gone down from an attack," McMahon says.

Fastest Government Web Site
This month the agency won kudos from Web metrics firm Gomez Inc, which said the IRS operates the fastest web site of all U.S. government agencies, and one of the fastest on the Web. The annual Gomez Web Performance Awards found that the IRS.gov web site had uptime of 99.97 percent and average response time of 0.56 seconds, less than half the time needed to load the WhiteHouse.gov home page.

The IRS data centers may play a role in the Obama administrations ambitions to shift some government applications to a could computing model. The agency has considerable data center space available as a result of its consolidation.

"There's a big drive for Treaury agencies to take advantage of existing space," said McMahon. "These are large data centers. The government wants to have its own cloud, and (IRS data centers) are being looked at for that."