What is our primary use case?
We are old school, we don't have proper documentation on how our application interacts with the downstream components. We tried CA Wily, that didn't provide a solution. We had CA Wily for three years, but when we deployed AppMon it gave us the complete picture, how the application interacts with downstream components, the application flow.
Based on that, my development team is developing the application automation diagram using the app map. It's like reverse engineering I guess.
How has it helped my organization?
It's really helping out. The term we call the mean time to repair, that has improved a lot. Before we had a "command center" type of setup. We had all the dashboards on the screen and five to six people sitting in front of the computer. When there was an issue they used to call a "bridge" meeting and everybody would login to the bridge.
The AppMon solution helped the OCC, the operations guys, to pinpoint one problem area and to call the specific group, instead of opening the bridge for everyone and asking them to join there.
The mean time to repair and the resource utilization time, they are totally reduced deploying APM.
What is most valuable?
The Transaction Flow diagram and the class and meta level information, those are really key selling points in the automation for AppMon. Also, the meta level instrumentation and the dashboards that most of the people use in our organization.
Most importantly the back-end components. Most of the back-end components that the application connects to; nobody knows how our application interacts with, for example, the DataPower Gateway. But AppMon really provides that information for us. So finding the gaps is the key here.
What needs improvement?
One thing I'm missing and that is the JMX from the MBeans, it's missing completely.
Also the reporting. We have a load testing team, they completely rely on the reporting for analyzing the data. They should have a template to create a report and they should have something to auto-deliver the report into your email box.
They also need to develop how to capture the JDBC and MBeans metrics. That is something they're lacking. Also integration for the extension to DataPower and MuleSoft Gateway.
The other feature that Dynatrace should have is - from what I see in Dynatrace in our PoC - when you auto-upgrade the agents, the JVM or the application has to be restarted. But if you have something like an "auto-attach" feature, to attach the agent for the running process, it would not require a JVM restart. That would be nicer. That is a killer point.
What do I think about the stability of the solution?
We haven't seen it that much but it's promising. We haven't had any downtime.
What do I think about the scalability of the solution?
We have a small portion deployed right now, but they're planning to go enterprise-wide, we're thinking about 8,000 to 10,000 agents, so probably at that time we'll know the scalability. But the small one we have right now, it's good. It's doing well. The UI is faster, people like it.
How is customer service and technical support?
We used tech support one time. We did in Fabric. We have internal cloud, called C3. So C3 and Fabric, we had a little bit of an issue, we called the support guys to fix it so and they were knowledgeable.
Which solutions did we use previously?
I have used HPE Diagnostics, CA Wily. I'm now doing a PoC with Dynatrace and AppDynamics.
With CA Wily, the problem was the complexity of our environment. It's not the tool, we have too many security layers. CA Wily, there's a breaking point. When we have the agents on multiple servers, it's not creating that link because of the security. But Dynatrace, I don't know how they did it, it's really awesome.
We switched because of the Transaction Flow diagram. There is a complexity in CA Wily that it is not able to integrate into our security layer, whereas AppMon can. That's the main thing that's missing. We were looking for that flow diagram. Nobody knows it. The developers call it but they really don't know what the main functionality is that's behind it: How the Apache layer connects to application server, and the application server connects to the different services, and the services to the back. Nobody knows it unless you have proper documentation. Proper documentation is very difficult to get in any organization.
Upper management really liked the Transaction Flow diagram. No matter what, that's the key.
Comparing AppMon to Dynatrace, I like Dynatrace very much because of the ease of use. AppMon is complexity. You need to know more to use the tool. The adaptability comes when you have a nice GUI that is easy to navigate. I saw that in the Dynatrace SaaS model.
What other advice do I have?
When I'm looking at AI, the problem identification and anomaly prediction is important. It's good to know, beforehand, when the problem is going to happen. The anomaly detection is the key area, and part of the AI I think.
If we had a solution that gives you an alert that said, "This is your problem, this is how you're going to solve it," that would be really awesome. Pointing out a problem to a specific group is a key point. That would really help, instead of globally alerting everybody, alerting upper management. If before they know it, you can solve the problem that would be nice.
When looking at vendors, we have a key set of requirements. Among them are container health monitoring, flow diagram. Also extension monitoring the non-Java applications, or non-supported applications, because Dynatrace works mostly on the .NET or Java applications. There are applications out there which are non-Java based like PeopleSoft. At least we can see the interaction with those components, but it would be nice to see what's going on inside those external components. That's what I'm looking in future releases, more support for things like PeopleSoft.
I would rate Dynatrace an eight out of 10 compared to other tools. The amount of data it provides is awesome. Other tools work on a sampling methodology but Dynatrace captures all the RAM sessions that are running. It's more data, but they have the filtering options so I can pick the data I want. Capturing all the data gives me more insight into what's going on, I can compare a bad response to a fast response, that type of thing. Capturing all the data is awesome.
I've worked with HPE Diagnostics, I've worked with CA Wily, now I'm doing a PoC with Dynatrace and AppDynamics. Compared to all these products Dynatrace stood out.
Disclosure: I am a real user, and this review is based on my own experience and opinions.
Feb 21 2018