Despite organizations pursuing multi-cloud strategies to ensure business continuity, resilience and performance, research conducted by Turbonomic and Verizon found that cloud purchasing decisions are still almost entirely focused on pricing ...
In Part Three of APMdigest's exclusive interview,
Jonah Kowall, Research Director in Gartner's IT Operations Research group, discusses additional new requirements for the Magic Quadrant for APM, as well as the challenges of application performance management.
APM: You mentioned that Gartner is requiring companies in the Magic Quadrant on APM will now be required to offer a SaaS option. In addition, you are also requiring vendors to offer all of Gartner's Five Dimensions of APM, to be included in the Magic Quadrant for APM, correct?
JK: Yes. We previously required 3 out of 5 dimensions, but this year to be included in the Magic Quadrant a vendor must have a full-featured APM product.
APM: Do you expect this requirement to cut out many of the smaller players in the market?
So we are starting to see many of these small vendors with one product and one agent be able to do all five dimensions. Since the last Magic Quadrant, about a year ago, we have seen many vendors that had four dimensions move to five dimensions, and those are all unified products.
APM: In your experience what are the main reasons why APM deployments fail to deliver?
JK: The number one issue that I see is a vendor will convince the customer that they should buy APM for all of their infrastructure and then they should work on deploying that and have a services engagement to do so. The issue is that when you buy hundreds of thousands of licenses of APM products, implementing those licenses across potentially hundreds of thousands of applications is not feasible in any reasonable amount of time.
Taking a much more balanced approach to identifying and implementing the critical components that should have APM on them tends to work better. Although the vendors do not appreciate it when you do smaller transactions with them, that tends to be the most effective use, and the reason why APM does not often deliver the value it is supposed to deliver.
The other challenge is that a company may not have the expertise in a tool that is potentially complex to implement and use. That is being combated by bringing APM downmarket and making it easier to use. Then the product can be leveraged not just by the third line developers but also the first and second line developers, in terms of being able to understand an application issue, and some type of diagnosis as to what the issue is, and who the issue should be escalated to. I think a lot of that is being driven by the push for simplicity versus the push for comprehensive solutions.
APM: How do you perceive Big Data impacting APM in the near future?
JK: There are two different ways to look at Big Data and APM. There is the Big Data that APM tools generate, and there is the ability to monitor and manage the applications that house Big Data infrastructure.
First, instrumenting and monitoring every transaction generates huge amounts of data. So we are seeing many APM vendors starting to implement Big Data stores themselves, and we are seeing that penetrate more and more of the market. And there are a few vendors currently undergoing re-architectures, moving from a traditional data store to a hybrid data store with a relational database on the front and Big Data behind.
Second is monitoring Big Data infrastructure. Most of the technologies are written in Java, which makes them a natural fit for an APM product. We have seen a couple of vendors start to message and market towards those who are running the data infrastructure, about ensuring performance. That is going to continue, and we will see more and more vendors position themselves as being able to monitor the data infrastructure.
The one area that has not been addressed by the market – which is not necessarily APM but is part of it – is data information assurance. How do you ensure that the data you are writing to your Big Data store is actually writing effectively? So that market is going to have to evolve as well. The Big Data administrator, or BDA, is going to need toolsets to be able to manage and monitor how the Big Data infrastructure is operating. That is something that has not really occurred at this point in the market. There has not been the same type of tools available for monitoring data infrastructure as there are for databases themselves.
APM: It seems like everyone is talking about Big Data this year. What do you see as the driver behind this interest?
JK: I think more and more companies are experimenting with Big Data infrastructure. There is obviously an influx of money that is going into that market, as far as the venture capitalists go. So I think part of it is being driven by vendors releasing products and venture capitalists funding those vendors.
And the other part is being driven by companies actually starting to experiment. There are use cases of critical transactional systems that are being run on the systems, which is kind of scary, in my opinion. A lot of the banks are starting to run pretty important transactions through these Big Data stores without actually having the proper tooling around them to understand data consistency. It is interesting what is happening and how the economics are changing the market.
APM: Do you have any predictions on how APM will evolve?
JK: We will have one or two predictions in the Magic Quadrant but nothing that I want to give away yet.
ABOUT Jonah Kowall
Jonah Kowall is a research director in Gartner's IT Operations Research group. He focuses on application performance monitoring (APM), event correlation and analysis (ECA), network management systems (NMSs), network performance management (NPM), network configuration and change management (NCCM), and general system and infrastructure monitoring technologies. Previously Kowall managed a global team of engineers and managers for MFG.com, and was responsible for monitoring and enterprise management software and architecture at Thomson Reuters.