Self-hosted instance maintenance - Part 1

[Gaurav Bajaj]

So far, we have covered the planning activities, design consideration as well as various teams required to set up and install a self-hosted instance of ServiceNow in previous blogs.

Now we need to uncover an even more important aspect of self-hosted instance, which is operation and maintenance of in this blog, Let’s talk about key activities for infrastructure management Support to run a high performing, stable and scalable platform.

Below activities are in addition to other application platform activities like reviewing scheduled job, long running transactions, performance dashboard etc. to monitor performance and application maintenance.

• SAR report for performance monitoring

A SAR report is a consolidated log of system performance metrics (CPU, memory, I/O, network, disk usage, etc.) captured at regular intervals. It helps administrators analyze server health, identify bottlenecks, and troubleshoot performance issues.

• CPU Utilization - %user, %system, %iowait, %idle
• Memory Usage - Free, used, cached, buffer memory
• Swap Statistics - Swap in/out rates
• I/O Statistics - Disk read/write throughput

It is recommended to enable SAR data collection on both application and database servers.

On CentOS and RHEL systems, this is available through the sysstat RPM package, which records a wide range of system performance metrics at regular intervals. You can define the frequency of data collection based on your organization policies.

 A key advantage of SAR reports is that they retain historical data for hours or even days, depending on configuration. This makes it easier to formulate a baseline of user load and performance stats over a period thus providing you with a trend analysis as well as being able to identify any performance bottlenecks. All this is going to be massively helpful during capacity planning exercise.

You can place running SAR report for both application and database server on weekly/monthly basis and have its part of checklist items to review performance and trends.

• Node sizing Validation

In a SaaS instance, ServiceNow takes care of scaling of application nodes based on load on the instance. However, it becomes customer’s responsibly to manage and maintain the required number of nodes to keep the platform stable without performance lags or delays.

You may probably start with 2-4 nodes initial installation however it depends on various factors – no. of concurrent users, no. of scheduled jobs running, product landscape – ITOM (discovery, service mapping), ITAM or the no. of integrations to truly define the number of nodes required.

After the initial installation, it’s important to revisit the node sizing regularly based on platform growth. This can be done by looking at SAR reports, performance loads or future new products and a new user starts logging into the platform.  

Continued in next blog....