System Resources

Monitoring CPU and memory usage facilitates efficient resource allocation and management within the BizTalk environment. 

Consider a scenario where the BizTalk environment orchestrates order processing workflows. During peak sales seasons or promotional events, there are often spikes in incoming orders, putting significant strain on system resources. By monitoring CPU and memory usage, administrators can identify performance bottlenecks and optimize resource allocation to ensure smooth operation during high-demand periods.

Through careful analysis of usage patterns, administrators can make informed decisions such as adjusting configurations, allocating additional resources, or optimizing orchestrations to enhance overall system performance. 

Setting up monitoring for System Resources

BizTalk360 automatically retrieves the current CPU and memory usage of the BizTalk server(s). By following below steps, you can configure BizTalk360, to alert you when the CPU and Memory go below the specified threshold percentages.

1. Go to Monitoring ->Manage Mapping ->BizTalk Servers ->System Resources.
2. Select the BizTalk server from the drop-down for which you want to set up the monitoring.
3. Choose the Alarm (see Manage Alarms) which you would like to associate the alerts.
4. Click the 'Enable Monitoring' button to set up monitoring for the system resources. The current memory and CPU utilization count is listed in this section.
5.  By default, the warning threshold limit is set to less than 40% and the error is set to less than 20%. The default threshold limit can be modified according to your business usage. 
6. Aggregation type - This option defines how monitoring should happen, whether based on the average value calculated during the last cycle or based on the count of threshold violations that occurred during the last cycle. 
7. Click 'Save' to save the threshold information.
8. At any point, to stop monitoring, click the 'Disable Monitoring' button.

What is Aggregation Type and how it works?

Aggregation is the process of collecting CPU and Memory usage samples for the specified time period and consolidating the result according to the chosen aggregation type.

In the Environment Settings -> Monitoring -> Mapping Configuration -> Threshold tab, users can configure the number of usage samples to be collected per minute. By default, the Maximum Polls Per Minute is set to 15 sec. This means that usage data is collected continuously for 15 seconds in a minute and that will be used for monitoring. If you want to monitor the resources usage for a longer period, you can increase this Maximum Polling limit.

BizTalk360 offers two types of Aggregation Average and Count.

1. Average: It calculates the average of the usage samples collected over the specified duration and compares it against the configured threshold rules. 

For e.g., suppose a user has set the Maximum Polls Per Minute to "20" and the Aggregation Type as "Average." In this case, the monitoring service will collect CPU and Memory Usage for 20 seconds during each cycle. For instance, if the current time is 11:30:00 AM, it will collect CPU and Memory usage from 11:30:00 AM to 11:30:20 AM, continuously for 20 seconds. The average of this 20-second usage data is then compared against the Warning and Error Thresholds.

In the above example, 

• A Warning notification will be sent if the average CPU consumption goes below 30% or average available Memory goes below 30%.
• An Error notification will be sent if the average CPU consumption goes below 10% or average available Memory goes below 10%. 

Users who typically prefer not to receive alerts for every short spikes in a minute can choose the "Average" aggregation type. This option sends alerts only if spikes occur frequently.

2. Count: The number of violations occurred in the collected usage samples determine the monitor result. 

For e.g. let's say a user has set the Maximum Polls Per Minute to "30" and the Aggregation Type as "Count". Hence the monitoring service will collect CPU and Memory Usage for 30 seconds during each cycle. For instance, if the current time is 11:30:00 AM, it will collect CPU and Memory usage from 11:30:00 AM to 11:30:30 AM, continuously for 30 seconds. 

When the Aggregation type is set to Count, users will be prompted for another option: Number of Violations. This option defines how many violations will be considered as an error or warning. If the value is set to 5, an alert will be triggered if any 5 collected samples in the 30 seconds duration violate the threshold value .

In the above example, 

• A Warning notification will be sent if any 5 collected samples of CPU or Memory goes below 30%.
• An Error notification will be sent if any 5 collected samples of CPU or Memory goes below 10%.

Users who prefer to receive alerts even if there is massive deviation between spikes can choose the "Count" aggregation type. This option considers the frequency of occurrences instead of overall average.