B2C Commerce Platform Monitoring Overview

Internal Monitoring
Commerce Cloud has built an internal monitoring and management system. This purpose-built tool is used by our Platform Operations team to ensure that our service is running properly and helps identify any problems before they become externally visible. This system is constantly being upgraded and developed by our Platform Engineering and Operations groups based on feedback from the teams managing the platform. This feedback loop within our group ensures that feature development addresses deficiencies that we experience while managing our service. 

We monitor each provisioned process every minute to ensure that it is responding to requests properly and within established norms. This monitoring mimics our load balancing schemes that we use to ensure that any failures that would alter our load balancing would be picked up. We provide notification to our platform groups of any issues ranging from informational (WARN level) to paging staff (FATAL level). This is all archived for trend analysis and used by our performance team to highlight potential inefficiencies within core applications (new releases), platform configurations, or customer implementations.

In addition to pass/fail monitoring we also collect usage information on a per process basis to measure efficiency. This helps us identify something like an uncached page that a customer deployed that could result in problems. Some of the metrics that we monitor and trend are:

• Requests per second (per instance and per process)
• Network IO (per instance and per process)
• Sessions per hour
• Database Usage per second (select/update/insert/delete)
• Object usage (per process)
• Memory usage (per process)
• CPU usage (per process and in aggregate)
• Cache efficiency (per process and in aggregate)

Note: These metrics cannot be provided to customers upon request.

External Monitoring
In addition to the exhaustive internal monitoring described above, Commerce Cloud validates that systems are available from three different network locations. This polling is done every 5 minutes from those three distinct locations in order to provide system availability information. This monitoring verifies that the platform is available and performing as expected. We do not test customer sites due to customizations that could cause our external monitoring systems to give our platform engineering group false positives.

This monitoring is the basis of our SLA. We use it to calculate the availability of each production instance in the realm. This ensures that your SLA covers what external consumers experience and not just the 'ideal' internal monitoring view of your availability. You can request your monthly report from your Salesforce support representative.


What can I as the customer monitor?
We encourage all customers to use an external monitoring company to validate their storefront availability. However since we can't control all possible network paths to our service we need to rely on our monitoring information for any SLA disputes. Since our external monitoring is provided by an external service we can't influence what availability it reports.

We have customers that simply poll their storefront, we have others that have written custom status pages that return the state of things like jobs, external integrations etc. So what you choose to monitor is completely up to your technical and business teams to decide. Our commitment to the customer is to ensure that we provide the resources that your storefront needs to make it available to the world and taking orders.