Getting more out of your application performance management (APM) tools means starting the process way before the purchase is made, according to some successful implementers.
“The biggest mistake you can make (with an APM deployment) is not to pay attention to the business process and the people,” advised Theo Gibson, director of infrastructure at Honeywell Aerospace based in Phoenix, Ariz.
Application performance management deals with monitoring the performance and availability of software applications.
Gibson and IT executives from American Express (Amex) and The Hartford Insurance Property and Casualty Company spoke about their experiences in deploying APM technologies at last week’s Symantec Vision 2007 conference in Las Vegas.
APM gives IT the tools to measure productivity by tracking performance and availability of business critical applications, explained Gibson. For instance, Honeywell’s APM tools enable Gibson’s team to quantify the effects of an application downtime or outage against business performance or productivity.
For American Express, which hosts a number of financial applications for stakeholders, APM allows its team to “understand time and quality” relative to service level agreements, said Michael McGibbney, director of technology infrastructure with Amex.
“We tie performance directly to availability,” said McGibbney, adding that his company also uses a dashboard product that enables his department to show the actual performance of business processes, which are then transparent to business partners.
Deploying the APM also allowed the Amex IT team to reduce support costs associated with servicing application outages, McGibbney said. As Amex uses standards from ITIL (IT Infrastructure Library) and Cobit for IT governance, McGibbney said the APM process was mapped back to these best practice standards.
“We turned from cowboy-style problem solving to a more efficient IT shop,” he said.
Doing an analysis and setting up the organizational processes involved with deploying the APM should be done at the very beginning, said Honeywell’s Gibson. This includes identifying the individual or group of individuals who will be responsible for driving the initiative, as well as establishing those who are going to be the system experts, he added.
Honeywell runs a mix of products for server and application performance management functions including management software from Mercury – which HP acquired last year – and Symantec Corp.’s i3.
It is also a good idea to assess the size and scope of the IT environment and doing all the necessary planning and analysis upfront prior to deployment, said John Lamb, assistant vice-president of platform management services at The Hartford.
This extensive preparation always comes in handy, especially when implementing such sophisticated toolsets as Symantec i3, Lamb said. “Symantec was our biggest challenge [because of the] sophistication of the tool…Doing all analyses upfront is the key. Without that, all sophisticated functions from [the technology] may hurt you,” said Lamb, explaining that prior to the i3 deployment his team was hugely reliant on spreadsheets for reporting and performance tracking.
For The Hartford, challenges around the APM implementation were threefold, according to Lamb. First is organizational challenge: “I had to create a framework that separated concerns to understand the different roles and responsibilities of each,” he said.
The next challenge was in planning the most effective way to get the tools rolled out across a variety of platform. The third was deploying the tools without adversely affecting the environment, Lamb said.
One of the more critical steps for Lamb’s team was integrating The Hartford’s legacy systems with newer architecture like SOA. To make that integration easier, Lamb’s team drafted an infrastructure sequence diagram that integrated Web applications with the mainframe environment, The Hartford executive said.
