Q&A: Inside Application Performance Management

Application performance management tools help you keep applications running at their best to help IT meet SLAs.

Keeping your business users satisfied means keeping systems up and running at their best. Application performance measurements are a key to maintaining your service-level agreements, but knowing what measurements to take is just the first step to troubleshooting problems and maintaining application up time.

To learn more about what application performance management tools can do, we turned to Rich Bentley, Compuware's director of solution marketing.

Enterprise Strategies: What is end-to-end application performance visibility? What benefits does it provide?

Rich Bentley: End-to-end application performance visibility is all about understanding what’s going on with applications and transactions across the entire infrastructure. This represents a more holistic approach to monitoring application performance and availability by looking top-down across the entire infrastructure, as opposed to the more traditional bottom-up, infrastructure-centric view of things.

This approach helps IT become more proactive in tackling performance or availability issues that may affect user productivity, customer satisfaction, or even bottom-line revenues in the case of customer-facing applications. It also helps IT solve issues faster and with fewer resources, which can result in significant cost savings within IT and an increased level of focus on strategic projects that can bring business value.

What are some examples of the types of measurement taken?

At a basic level, you want to look at round-trip application response times, but monitoring real user transactions from an end-to-end perspective allows us to also look between the different infrastructure tiers that support the application. Not only do we see which business transactions are performing slowly, we can also see which transactions may also be slowing down at individual tiers. This might be represented by things such as slow Websphere MQ transactions, slow Web services, slow Java or .NET calls that we have to dive into, or slow back-end database queries.

How is this information delivered, and who sees it?

There are several different audiences that benefit from this information. IT management uses real-time dashboards to become aware of service issues before end users complain. Then, network managers, server administrators, database administrators, and other technicians can dive into deeper analytics with drill-down reports that help pinpoint the root cause of problems. IT executives leverage high-level dashboards that tell them how IT is doing overall in terms of service-level quality, and then this information can be used to enable an effective dialog with business customers on how well IT is meeting service-level expectations.

Does application performance visibility help IT troubleshoot and resolve issues, or does it just make IT aware of issues?

It helps with both. On the troubleshooting and resolution side, having an end-to-end view makes it much easier to isolate the fault domain for a performance or availability issue. A common example that we see in organizations without end-to-end APM is that users are complaining about a problem, but the network and server monitoring tools are all showing green lights. So they end up getting a war room together with the best and brightest IT resources trying to figure out what’s causing the problem. These war rooms can drag on for days and may include ten or more people.

Having end-to-end visibility allows you to, first of all, verify that there is an issue, quantify exactly how bad the performance is, and determine how many users are affected at what locations. This starts to point IT in the right direction, but taking it a step further, IT can quickly see where transactions are slowing down in the infrastructure, and also correlate key performance metrics with the end-user experience to see what was going on at the time service degraded. For example, we may see that CPU utilization and response time on one of the Web servers started spiking around the same time that the end-user experience started going downhill. This isolates the problem so we can assign the appropriate server administrator to investigate further.

By doing this we can essentially eliminate the war room response scenario and solve problems faster. One company we worked with is now able to solve 80 percent of all application problems in one hour or less.

Are traditional network and server monitoring tools still needed?

Absolutely. IT still needs to monitor lower-level components to understand what’s going on at a deeper, technical level. It just shouldn’t be the starting point, and these tools definitely shouldn’t be relied on to provide metrics that are shared with the business customer.

End-to-end performance solutions can actually help companies get value from the investment they’ve already made in infrastructure monitoring tools. One problem IT has is that they get so many alerts and events happening at the component level and there’s no way to wade through that sea of information to get anything useful. By linking end-to-end monitoring with traditional monitoring tools, we can understand which of these events are actually related to or causing a poor end-user experience and then use the deep-dive analytics provided by these tools to troubleshoot and fix the problem.

Does this information help with service-level management?

Yes. One of the primary inhibitors for adopting service-level management is the difficulty in measuring the items that matter to the business. SLAs that are based on server uptime or network bandwidth are of limited use for ensuring business expectations are met. The business just doesn’t care about these things. They want to know that a service is available and performing at the right level, and that’s what end-to-end performance monitoring provides.

How would effective application performance management help IT align its activities with business goals?

One of the biggest hurdles to IT/business alignment is that IT speaks a different language than its business customers. While IT is generally very comfortable talking about bits and bytes, they are less comfortable talking about business services.

Having an end-to-end perspective that starts from an understanding of the end-user experience fundamentally changes the conversation. It allows IT to have a constructive dialog with the business based on real facts about service quality. This makes it easier to talk about realistic expectations and the value that IT is actually providing. For many of our customers, this has given IT credibility to the point where they are now viewed as a strategic partner to the business.

What impact do technologies such as virtualization have on application performance management?

Technologies such as virtualization, service-oriented architectures, and cloud computing hold great promise in helping IT become more agile and cost-effective in delivering the applications and services that drive the business. However, we pay a price for these new capabilities in the form of increased complexity and new management challenges. We now have many moving parts, highly distributed applications, and greater difficulty getting visibility into our infrastructure.

This really emphasizes the need for a top-down approach to service delivery, because the bottom-up approach simply won’t work anymore. Monitoring from the end-user perspective is really the only way to effectively understand how an application or service is working.

What mistakes does IT typically make when trying to manage application performance?

We’ve already talked about the most common mistake, which is thinking that traditional monitoring tools will give you what’s needed to effectively manage application performance. There are a few others we see regularly.

First, many IT organizations try to do everything at once. Projects that attempt to monitor every application and every transaction are doomed to failure. Start by having a conversation with the business and understand what are the most critical applications and transactions, and what is the impact if those were unavailable or performing poorly.

Second is the misconception that IT organizations can manage application problems by “throwing” hardware at the issue. This approach assumes that application performance problems are introduced by system faults and capacity issues. This turns out not to be an effective way of dealing with issues and results in a tremendous amount of wasted money. The only way to truly manage application performance is to look at the problem from the view of the application.

Finally, we see a lack of communication between the silos of IT causing problems in all areas, but especially in the application management discipline. This becomes apparent when changes are made by one team that may affect others, and it really highlights the need for standard change-management processes that are adopted across the organization.

What is the future of application performance management? How do you expect this technology to evolve?

In many ways, an application can be thought of as a proxy for a business service. For most IT organizations, monitoring an application is the closest they can get today to measuring and reporting on something that is directly relevant to business results. We view application performance management as a stepping stone to true business service management, and we see the technology evolving in this direction. The problem is that business service management is currently viewed as a giant leap for most organizations, with a full-blown CMDB as a prerequisite and expensive, standalone tools and consulting engagements required to get started.

We recently introduced technology in our Vantage 11 release that will make it much easier for organizations to adopt BSM. By monitoring all applications and all users at every tier within the infrastructure, we can start to build a service model with very little effort. This service model allows us to understand the interrelationship between a business service and the IT infrastructure that supports it, which is what true business service management is all about.

What application-performance products or services does Compuware offer?

Compuware Vantage is our primary application performance management solution. Vantage provides a top-down approach to ensuring service delivery that starts from the end-user experience. We deliver an end-to-end view of application performance so that IT can proactively identify and resolve problems before they impact the business.

On the mainframe side, our Strobe product has been the industry leader for a long time. Vantage and Strobe are tied together so that we provide an end-to-end view from the end user to the mainframe, which is especially important as so many applications span across both distributed and mainframe transaction systems.