Will AppDynamics (Performance Monitoring Tool) slow down my Production Application?

Question

Is it ok to deploy Performance Monitoring Tool - AppDynamics - in Production? The application to be monitored is a standard Java/J2EE Web Application.

I have never worked with AppDynamics, and my concern is that it may actually slow down my application.

Has anyone used AppDynamics in Production? Or should it be used only in Test kind of enivornments.

Crenate

Answer 1

Full Disclosure: I currently work for AppDynamics.

AppDynamics was designed from the ground up for high volume production environments but works equally well in both prod and non-prod. It's currently running in production in some of the worlds largest mission critical application environments at Netflix, Exact Target, Edmunds, and many others. Here are a few quotes from existing customers…

"It's like a profiler that you can run in production" -- Leonid Igolnik, Taleo

"We found that the overhead was negligible" -- Jacob Marcus, Care.com

"We wanted a monitoring solution that wouldn't impact our production runway" -- John Martin, Edmunds

AppDynamics overhead is extremely low but I suggest you test it and see for yourself. You can download and use it for free from the AppDynamics website. Good luck in your search for the right APM tool.

Aboutface

We are using AppDynamics at work, and not only it is not affecting our performance, it saved us to rollback a very important application identifying some critical performance points causing major outages on daily basis. – Appaloosa

Answer 2

There are a number of assumptions made by a vendor but the following are most common:

1. We assume your application request processing time is sufficiently high to dwarf our significant overhead.

TRANSLATION: We assume you have a slow performing database backend.

1. We assume that instrumentation is applied to a very limited section of the code source to lessen the impact of our significant overhead.

TRANSLATION: We assume that you already know your performance hotspots.

1. We assume that there is large amount of under utilized processing capacity to offload our significant overhead.

TRANSLATION: We assume that you will not notice tricks used to hide our overhead.

1. We assume that it is impossible to realistically and reliably measure our significant overhead.

TRANSLATION: We assume that you know little about performance engineering.

And my favorite (5) is the restriction within a vendors software license on the publication of benchmark results.

TRANSLATION: We assume that you blindly accept our claims – unquestionably.

Finis

Answer 3

Yes it will if the application is sensitive to extra GC cycles caused by call stack sampling. The impact will depend on the number of threads and typical call stack depth. This is not specific to AppDynamics, other call stack sampling solutions such as NewRelic and VisualVM Sampler will have a similar impact.

http://www.jinspired.com/wp-content/uploads/2011/11/JXInsightOpenCore_Good_APM_vs_AppDynamics_Bad_APM.pdf

http://www.jinspired.com/wp-content/uploads/2011/11/JXInsightOpenCore_vs_AppDynamics.pdf

Finis

Answer 4

Appdynamics will not slow down your system significant, I was on a usermeeting and they said that they always try to be under 2% cpu usage, thats nothing compared to what you get from them. They are working with samples per time, so if you have 10 requests per second or 100, they will still take the same amout of your cpu / bandwith / whatever.

Hypochondriasis

Answer 5

I suggest to look into Gartners magic quadrant and get dynaTrace since it has negligible overhead , less than 1% in production under load.

Dineen

I am sorry but there is not a shred of truth to that claim. In fact dynaTrace has a very significant impact because of its lack of adaptiveness in instrumentation and measurement as well as the fact it collects far too much state diagnostics in the most expensive way imaginable. – Finis