CTO at a computer software company with 51-200 employees
Vendor
2014-10-06T05:05:11Z
Oct 6, 2014
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
Search for a product comparison in Application Performance Monitoring (APM) and Observability
System Engineer at a aerospace/defense firm with 1,001-5,000 employees
Vendor
2014-10-03T14:29:52Z
Oct 3, 2014
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..
Solutions Architect at a comms service provider with 501-1,000 employees
Vendor
2014-10-03T08:18:07Z
Oct 3, 2014
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.
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.
Also this book (www.amazon.com) 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.
Find out what your peers are saying about Datadog, Dynatrace, New Relic and others in Application Performance Monitoring (APM) and Observability. Updated: April 2024.
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
CTO, COO and Senior Engineer at a tech services company with 51-200 employees
Consultant
2014-10-02T11:59:02Z
Oct 2, 2014
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.
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
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
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.
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.
Also this book (www.amazon.com) 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.
This is good guide
www.researchinaction.de
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
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
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