In APMdigest's exclusive interview, Bojan Simic, President & Principal Analyst, TRAC Research, shares the APM blind spots and challenges revealed by his new interactive report on Application Performance Management technologies and the market, which will be introduced this month.

Click here to start with Part One of APMdigest's exclusive interview with Bojan Simic from TRAC Research

APM: What are the major differences between your survey results this year and last year?

Simic: One is the major concern about management cost. This year we are seeing more people saying that is a major challenge. You can spend a couple of hours instrumenting the tool every time the application changes. Using an agile methodology, you are making changes to your code on a weekly and sometimes daily basis. If you have to go back and instrument your APM tools every time you make a change, that becomes a major mess. At the same time, many application performance management tools are not adjusted to follow that, and it requires a lot of additional work.

APM is no longer something that is only appealing to large enterprises that can afford to throw a couple full-time engineers at the management tools. More and more companies need APM, but not everybody can allocate the resources in-house to make sure APM works on a daily basis.

The other major difference is that organizations require more visibility into the web browser. That has a lot to do with how applications are being designed these days. Your typical web application today has a lot of content and third-party services, and it depends on components that your organization may not be able to control. That drives more need to be able to see inside the browser.

Taking a more agnostic approach for monitoring the quality of the end-user experience also requires greater visibility in the browser. If you are monitoring points on your infrastructure, they won't be able to monitor the cloud applications. So you need application monitoring that is agnostic, so you don't really care where you are hosting these applications. Having more visibility within the browser makes this monitoring approach more agnostic of where the services are.

APM: What are the main APM blind spots users are experiencing?

Simic: We defined “blind spots” as when the user does not see the element of the delivery chain that actually impacts the quality of the end-user experience.

Number one was the web browser. There is definitely more processing that is happening close to the end-user on the browser or on the user device. As opposed to a couple years ago, there is more dependency on how browsers can process this information. But monitoring network traffic, database and server does not provide any visibility into how the browser impacts the quality of the user experience.

Another blind spot is the virtual machine. This is interesting because one of the major promises of virtualization technologies was that they will help break down some of the silos within IT. What ended up happening was that virtualization projects actually created another management silo.

The majority of virtualization management tools are focused on capacity planning, utilization and availability metrics, and very few of them actually have capabilities that will tell you how it will impact the user experience if you change something in the virtualized environment. Many survey participants reported that they are getting some measurable benefits on virtualization, but the correlation between the performance of virtual machines and the quality of the end-user experience is a blind spot.

Mainframe is another interesting blind spot. For applications that span across mainframe and distributed systems, they lose a lot of visibility because of the way the monitoring agents are set up, the way the whole monitoring capabilities in general are defined, they are actually not as effective monitoring transactions as they cross the mainframe.

APM: How can these blind spots be addressed?

Simic: There are technologies out there that address these blind spots. For example, there are technologies for monitoring the performance of virtual machines or the mainframe. But in many cases these tools only monitor virtual machines or the mainframe. They have no APM capabilities to show the impact. When you move this virtual machine from one physical host to another, what is going to happen to the quality of the end-user experience? They have no visibility into how their actions are impacting application performance.

One solution may be for APM tools to integrate better with the tools that are monitoring some of these blind spots.

The more effective approach will probably be to deploy APM capabilities to be able to track transactions across different areas of the application delivery chain.

APM: What are the key challenges people are experiencing with APM technology?

Simic: There are three high-level challenges for APM. First, many of people reported that they are able to collect more data, and they are able to see some areas of APM that they have not been able to see in the past, but they are still spending a lot of time in “war room” meetings, troubleshooting problems -- and there is still a lot of finger-pointing going on.

If you can reduce time being spent correlating and normalizing data from different sources, that will definitely help address that problem. We are seeing a trend in this market that, in addition to getting APM tools for collecting data, organizations are also investing in analytics that stand on top of that. They are trying to make sense of the different types of information. With analytics, they can find exactly what they need, in the time they need to fix the problem, as opposed to having to filter through tens of thousands of data points that are all valid but may not be relevant to the problem.

The second challenge is about being more proactive. Being able to see performance issues or potential problems before users start calling the help desk or filing trouble tickets. As mentioned in the previous challenge, organizations are looking to repair problems in a timely manner, but even more important to them is being able to learn about potential problems before end-users.

A third challenge is the management cost. For some APM tools, the configuration or instrumentation process is very time-consuming and resource intensive. Trying to find the balance between how much time they can afford to spend on managing not just the application but the management tool versus how much visibility they can get. They need to find the right balance.

Making the deployment process easier, the whole process of dealing with application changes, code changes, adding new applications -- if vendors can figure that out, it can open up a lot of opportunities for them. For many user organizations, especially in the midmarket, the perception of APM tools is great, they see the value, but they can't afford it, they don't have resources in house to manage these tools, because they are not easy to deal with.

APM: You mentioned analytics before. Do you see analytics addressing many of these challenges?

Simic: Definitely. That is becoming one of the major differentiators of APM products. You also have vendors that are not necessarily pure APM vendors, but they specialize in analytics. Deploying those types of solutions on top of more traditional APM products definitely provides value.

I think your point is more than valid. Look at the first two challenges of APM: time spent troubleshooting problems and being more proactive to solve issues before users are impacted. Analytics can play a major role in both of those challenges.

APM: Where exactly does analytics fit into APM?

Simic: Analytics definitely cuts across all of the APM capabilities. But also, at the same time, we have an important group of vendors that offer analytics, but don't fit into any of the three categories of APM, such as Netuitive. Analytics is almost becoming a fourth category in APM. That might become a fourth segment of our study down the road.

APM: Is automation one of the primary solutions to the management issue?

Simic: It is. In terms of troubleshooting, organizations are becoming more open to automating some of the troubleshooting processes, but they are also being very cautious about how it impacts other areas of their organization.

If an organization is not 100% confident that the information they are getting is accurate, ii is very difficult for them to build automation policies on top of that. Unless you get really comfortable with the accuracy of the information, you don't want to set something like that up, because it can mess up a lot of the things that you do not want to touch. Organizations are selective when it comes to how automation is applied to APM.

But also, at the same time, when it comes to management cost, automating some of the configuration policies and reducing the manual work is definitely something that appeals to organizations.

APM: What key capabilities do you see driving adoption of new APM tools?

Simic: One of the driving factors behind APM adoption are capabilities for correlating information collected by different types of tools. Companies are using one product for user monitoring, and a different product for synthetic monitoring. They might use a different product for Java applications. They're looking at a number of different screens and they can't make sense of all the data that they have. So being able to actually put all this information in the same place where it can be analyzed together is the type of capability that organizations are saying they can make budget available, if there is a tool that can do that.

Also, the ability to adopt transaction-centric approach for monitoring different aspects of application performance is becoming increasingly important. This capability allows organizations to have a unified view into different elements of the application delivery chain that are impacting the quality of user experience.

Click here to read Part One of APMdigest's exclusive interview with Bojan Simic from TRAC Research