More and more NAV servers are being deployed to the cloud, yet a large number of NAV customers are keeping their NAV installations on premise (either on a physical server or a VM). Those that choose cloud(and if you didn’t please read this) get the benefit of cloud specific tools for measuring the hardware and database performance.
Unless you have a specialized tool, to monitor the performance of your server you need to establish a baseline. During server’s deployment, administrators can still tweak (especially if it’s a VM or part of a hyper-v infrastructure) the server hardware until the server reaches a satisfactory performance level.
Regular recordings of server’s performance can uncover problems in their early stages, sometimes long before they become obvious in production.
Windows Server 2008 and later server versions as well as Windows 8/8.1/10 come with two easy to use tools for measuring (and scheduling) the performance of the system: perfmon and logman.
With perfmon we get a graphical interface in which we can manually create our alerts or counters, run them and analyze the results.
With logman we can manage the counters from command line.
On my system I created two counters, one for measuring SQL server counters, the other was targeted at the hardware performance:
Double-clicking on HardwareBaseline we can manage the counters:
To create these two counters I ran the following script:
To start and stop the counters, run:
Or manually, in Performance Monitor, right click on the counter and choose Start or Stop.
A few “logman” command switches I use:
-b and -e switches to allow the performance counters to run in a specific time period.
-v switch to add a timestamp for when the output file is created
-f specifies the file format used for collecting data
After a few minutes the performance counters graph will look like this:
With a Task Scheduler entry you can control when to start and stop the performance counters collection.
As for the analysis of collected data, there are lots of places online where you can find valuable information on counter’s results interpretation.
Download the package with hardware and SQL counters and a sample script from here.
A baseline case study
One of the processes that is pressing all resources on a NAV server instance in our solution is the running of a report that posts sales invoices for all customers for a specific Due Date. I’ll run the report and record and discuss the counters recorded.
With the installation of MS Dynamics server you get out of the box a few counters NAV specific:
I created a new data collector with the following counters:
In Performance Monitor right click on your new Data Collector Set and Start.
Next I went to RTC and ran the report.
After the report finished I came back to Performance monitor and stopped the Data Collector Set.
Microsoft Dynamics NAV\.NET CLR\% Time in GC:
If RAM is insufficient for MS Dynamics NAV Server instance, you might see a spike in the “% Time in GC” – which measures the .NET memory garbage collection activity. My process shows a maximum of 7% nd an average of a bit over 2% – numbers that do not show that NAV is looking for more RAM.
Microsoft Dynamics NAV\% Primary key cache hit rate:
This counter should be above 90% – otherwise your cache might be too small because either your cache settings are set too low or the cache is shared between tenants and might need to be increased. I my case is above 99%:
% Calculated fields cache hit rate averages 63% which means that in 63% of the cases when we launched a CALCFIELDS command we hit the cache – decent number!
# Open Connections is 5, and refers to the number of open connections made from the service tier to the database hosted on SQL Server. You might be interested in the counter “# Active Sessions” which keeps track of the number of concurrent connections the service tier manages.
The rest of the counters give numbers of rows, which might or might not be too relevant considering that in time the number of rows increases.
Having a baseline and comparing regular performance counters log against it, is not just a proactive measure in order to keep a NAV server healthy. It is fairly easy to use and cheap (logman and perfmon are both builtin Windows), qualities that appeal to NAV customers and Dynamics VARs.