It has become increasingly important that applications, which many companies rely on to drive revenue, meet users' expectations. Unfortunately, the complexity of today's application architectures makes troubleshooting performance issues difficult at best. IT teams need to quickly identify the root cause of problems -- preferably, before the issues affect users. To do that, they need an application performance monitoring (APM) tool.
"Back in the day, I'd open one server closet, and it had to be there in one of those machines," said Archie Roboostoff, software director at Micro Focus, a software provider based in Rockville, Maryland. "Today, it's more difficult to find the root cause due to the fact that there's all kinds of easy-to-consume services all over the place. Hence the reason monitoring is critical."
Dave Link, co-founder and CEO at ScienceLogic, agrees. ScienceLogic is a provider of IT operations management software. Traditionally, users would log in to and interact with applications that resided in an on-premises data center, he said. IT managed the infrastructure and generally had access to the network and servers.
"Now there are benefits to moving resources for applications that would traditionally be managed in a private data center to cloud computing platforms. [But] IT loses some control because they can't walk down the hall, pull something out of the rack, check the connections, etc.," Link explained. "As we move to the cloud, we're relying on the resilient architecture and the infrastructure in these extraordinarily large data centers to provide compute as a service."
Shane Sheltonsenior director of application performance and DevOps, McGraw-Hill
For Link, this shift is changing the way program managers do their jobs and the types of tools that they need for a comprehensive view of their applications.
To further complicate matters, there isn't a standard for heterogeneous technology vendors, Link said. Cisco has its own proprietary application programming interfaces; Amazon has its own; VMware has its own. "They want you to stay in their camp, but it's not practical or all that common," Link said. Without a standard, it's tougher now for IT to maintain a clear line of sight into services from disparate vendors.
Dave Ursprung, senior IT architect for Catholic Health East and Trinity Health in Livonia, Michigan, knows this problem well. Almost everything in the health care organization's IT environment is virtualized, and visibility was a problem before deploying an APM tool. "We were using some of the raw tools that were provided by the vendors, and those tools were cryptic at best," Ursprung said. His team chose APM tools from Riverbed Technology, which allows them to "see pretty much everything in the virtualized environment."
This visibility enables Ursprung and his team to quickly solve performance issues. Before implementing APM tools, getting a system back online was quite difficult. Ursprung said it would require communication with 20 people, and each employee would be responsible for a different piece of the application architecture. It could take days to determine the source of the problem, he said.
Now, Ursprung receives alerts before performance problems affect end users. "It takes a lot less manpower and a lot less time because we don't have to pull together all those people," he said. Ursprung now has greater success pinpointing the problem and working only with the teams involved in the root cause.
Making transitions work
Shane Shelton is the senior director of application performance and DevOps at McGraw-Hill in Boston, which also uses an APM tool in its production environment for operational support. Shelton explained that McGraw-Hill is transitioning into a position as a digital software company, which makes application performance critical.
"We're talking about our entire revenue stream. So if the product is down, the company is not making money," Shelton said. His team chose Compuware's dynaTrace APM tools as a means of detecting root causes, triaging performance issues and fixing systems faster.
Ideally, performance problems are addressed before they affect users. To this end, IT teams are trying to get further ahead of potential problems. Ursprung and Shelton use APM tools in their testing environments to solve some of those issues before applications go into production.
"When we're about to release a major build into production, we will run a performance test on the environment in the data center or in the cloud," Shelton said. "DynaTrace will monitor it and give data back if there's an issue with a build," he said.
Joseph Dissmeyer, a programmer analyst for a large, international media company that he's not authorized to identify, uses SolarWinds' Server and Application Monitor to test performance in custom-built applications and third-party applications. "We will get a version of the software we want to roll out company-wide and do testing first, with monitoring set up. If there is anything abnormal, we send it to the software vendor and tell them we won't deploy their software until the problem is solved," Dissmeyer said.
Not only has this due diligence paid off in terms of delivering better-performing applications to end users, it has transformed IT. Performance monitoring is no longer about testing applications, pushing them to production and finding out later that there's a performance problem, Dissmeyer said.
When a problem pops up, Dissmeyer and his team not only fix the problem, they also ask themselves, "Are we monitoring it? How can we make sure it's being monitored?" He said, "[APM] has completely changed how we think about support."
APM tools: A must-have for the cloud
Complex apps need network-based APM tools