2014-10-02 11:21:00 UTC

What are the must-have criteria for an APM Solution for a large enterprise?


Is the single most important metric when measuring applications performance response time?

Guest
99 Answers
author avatar
User

Um... Usually response time is key to APM including request transmission and/or round trip time.

And following criteria should be considered;
- System Resources Usage (Processors, Memory, Disk, Network)
- Operating System Activities (fork, exec) that may affect to system performance

DJ

2014-10-06 05:05:11 UTC
author avatar
Real User

In my opinion, yes, average response time is definitely one of most important criteria for monitoring application
performance. Besides for providing pro-actively monitoring solution, it's
important for APM tools to setup application performance baseline
automatically. For example, the tool has the AI functionality to learn your
application behavior, which can set up baseline based on the performance
data it collects..

Zhang

2014-10-03 14:29:52 UTC
author avatar
Real User

Besides all the wonderful things APM tools deliver today, I personally think these add value.
• Having the ability to measure all transactions and report the good, the bad and the ugly, not just the ugly.
• Understanding the transaction path and then having the ability to report/trend transaction time spend at each segment (firewall, network, web application, database, user, etc.)
• Transaction tracing/discovery.
• Trapping the payload when things go south.
• Flexible/customizable dashboards.

2014-10-03 08:18:07 UTC
author avatar
TOP 20Real User

Must have criteria for an APM solution for a large enterprise is a simple question but the answer can certainly get quite convoluted.

Essentially when looking at application performance response time (across say key applications) the metric that I try to educate my customer base with on our solution set (Fluke Networks TruView) is EURT or End User Response Time. This is made up of application, network (composing of DTT, NRT), client and server.

Instantly if your dashboard or monitoring solution is detailing this you know exactly where to start your analysis. This simple triage screen can save hours of time. Additionally, as is the case with TruView, having Flow data and SNMP all together can then very quickly help with a network issue if the EURT composition is pointing to metrics such as data transmit time (DTT) or network round trip time (NRT).

There are then a number of other steps for additional analysis, but I have seen a majority of issues identified from a single screen of data.

Answers can also vary for the types of teams using APM. APM covers a huge array of areas these days from packet based data, flow and SNMP data, agent based monitoring such as .NET/Java/PHP and synthetic transactions.

Knowing the exact makeup of the environment for monitoring is almost as important as this simple question posed. I have just answered based on AANPM (or NMPD if you want the Gartner market) for providing an all-round solution set covering network and application.

2014-10-03 02:56:24 UTC
author avatar
Vendor

Also this book (http://www.amazon.com/APM-Best-Practices-Application-Professionals/dp/1430231416) describes best practices for APM including assessing your current monitoring stratgy and planning a POC for an APM solution.

As Brian said for a large enterprise scalability, capacity and broad coverage (from end-user/mobile through several layers of application servers and middleware probably onto the mainframe) are key.

Process, communication and collaboration is key to a successful APM initiative as it crosses the traditional silo boundaries in IT and brings Dev and Ops closer together with a common/shared view. Therefore early integration into the SDLC is important. The most mature/successful enterprise hardly need to use APM in production (any more) because they detect performance issues earlier in the SDLC and have performance related Go-Live criteria.

2014-10-02 20:35:45 UTC
author avatar
Vendor

For a large entreprise, the most important would be to focus on the essential and relevant indicators. These indicators are response time for transactions and network, errors that occurs. Then efficient and easy way to investigate inside the application. this is exactly what we try to do with Nudge-APM, keep it simple and easy to use, focus on main functionalities only.
Nguyen

2014-10-02 14:58:57 UTC
author avatar
TOP 10Real User

What are the must-have criteria for an APM Solution for a large enterprise?

Some would be:

· Consider Performance early, and throughout, your SDLC(s)

· Load/performance test early, and throughout, your SDLC(s)

· Monitor, analyze and strive for as deep of RCA as possible throughout levels of testing and in Production

Is the single most important metric when measuring applications performance response time?
While important, others would be:

· Scalability

· Capacity

· End user experience

· Functionality – functionality at low usage volume can differ when volume increases

2014-10-02 13:44:24 UTC
author avatar
Real User

The 3 metrics to watch would be Application Response Time, Network Response Time and Data Transfer Time. That combined with Packet Capture and Flow statistics will provide a very good picture of overall infrastructure health related to application performance.

Hope this helps.

Paul Jasina

2014-10-02 11:59:02 UTC
Find out what your peers are saying about Dynatrace, AppDynamics, Datadog and others in Application Performance Management (APM). Updated: November 2019.
382,745 professionals have used our research since 2012.
Sign Up with Email