NAV TechDays 2014 - Monitoring & Diagnosing Microsoft ...

Administrator

NAV TechDays 2014 - Monitoring & Diagnosing Microsoft Dynamics NAV Server Perf..
Presentation during the NAV TechDays 2014 conference of 20 & 21 November 2014 in Antwerp (Belgium)

Presenters: Dmytro Sitnik, David Worthington

Presented on Thursday 20 November 2014, 14:00 - 15:30

http://www.mibuso.com/dlinfo.asp?FileID=1608

Discuss this download here.

MatthiasKönig

Hello,

Dmytro Sitnik shows in this session a small c# tool (the "Noisy Neighbor" sheet in the presentation). Did anybody knows if I can find a download of this anywhere? on codeplex I does'nt find it.
Or if Dmytro Sitnik reads this, can you send us an download link? that would be great!

Kind regards,
Matthias König

wortho

Hi Matthias,

You can find the source code here:

https://navappprofiler.codeplex.com/Sou ... y/Form1.cs

Regards,

David

wortho

Hi,

The presentation was a great opportunity to see how NAV developers are interested in tools to analyze Dynamics NAV performance. There were also a lot of tough questions afterwards :-)

What I learnt is that having a few performance counters and the infrastructure for diagnosing NAV performance is not enough. The “proper” way to create benchmarks for a system is to collect data from the performance counters over a statistically significant period of “normal” usage and then analyze the data to define the expected metrics for normal usage. This is time consuming and I think most NAV developers would like to have an easier way of determining whether their system is performing normally and to quickly identify any obvious issues.

There are tools for analyzing SQL server performance that provide a pre-defined set of “normal” performance metrics and configuration parameters to identify any obvious issues. This is not a comprehensive performance analysis but it makes a good starting point for investigations.

What is needed for NAV is a set of guidelines for NAV performance counters and configuration parameters. I will be collecting data from my own investigations to build a set of guidelines.

Let me know if you have any comments or suggestions.

P.S. I have started a new blog to share my thought about NAV Performance here http://navperformance.wordpress.com/

kriki

wortho wrote:

There are tools for analyzing SQL server performance that provide a pre-defined set of “normal” performance metrics and configuration parameters to identify any obvious issues. This is not a comprehensive performance analysis but it makes a good starting point for investigations.

It is just to start with. It does not mean that when you have all you metrics better than this set, you have good performance. In SQL Server it is always : It depends!
It depends on hardware, number of users, how much they are working,...............

So, the ONLY real set that works is this: you need to create a baseline of performance metrics when all works well on your system. And you also need to create new baselines on a regular basis because your baseline may change and also to see trends.
And when you have performance problems, you can confront it with the baseline to determine where the problem is.
And when you have some trends, you might also see that somewhere in the future you might run into performance problems and act pro-active instead of re-active.
But all this takes time to define, to setup, to create and update and check the trends.
Something that is quite difficult to sell with our type of business (lots of small and medium-size customers/projects).

wortho

Hi Alain,

I think we agree. My point is that it may help developers to get started with monitoring if they had some generic numbers to begin with.
Some examples might be:
% Primary key cache hit rate > 90%
Heartbeat time (ms) < 10 ms
Average server operation time (ms) < 100 ms

I agree they might not be the right numbers in every case but at least they give the developer some hints about what "normal" is.

Regards,

David