Do you think it's important for a web property to have the following?
- Ability to detect application outages from a customers perspective for both web apps and APIs
- An accurate uptime score, based on if the application is Up for the customer
- Ability to be alerted to user experience degradation
- Ability to diagnose and troubleshoot problems with CDN or DNS
- Ability to show performance gains over time from a customers perspective
- Ability to track performance gains by introducing changes to the CDN
- Have a public status page that reflects the operational status of components from a customers perspective
If so then APM and other origin based monitoring is not enough. You need Synthetic and Real User Monitoring (RUM), New Relic calls this Browser monitoring, in addition to APM to achieve the above. The Red Hat Customer Portal did have an old synthetic monitoring platform, but it had many shortcomings that limited it's usefulness such as a proprietary scripting language. In 2015 a cross departmental team with members from System Engineering, Subscriber Platform, PlatOps and IT-Pnt, evaluated several Synthetic monitoring tools from different vendors to fill this need.
From that evaluation New Relic out classed the competition on every point.
How we are using Synthetics and RUM today
Since we chose New Relic for our Synthetic and RUM monitoring tool we have been using it extensively on the Customer Portal with great success. We currently have over 50 New Relic Synthetic monitors + their RUM product (Browser) monitoring the Red Hat Customer Portal applications such as:
- Case Managment
It gives us:
- Outage detection from customers view
- Alerts when there is an outage that affects customers even when origin server monitors are fine
- Accurate Uptime Score
- Insight into users experience as they browse
- Diagnose and find and troubleshoot problems related to the CDN and DNS
- Show performance changes over time
- Pre-prod and internal monitoring with Private Locations
- Automated public status page status.redhat.com
And we continue to expand adding more monitors as we identify parts of the portal that need it. As well as looking into consolidating APM and Synthetics in New Relic under one Alert system.
Our Monitoring Architecture
Example of a Synthetic monitor detecting a DNS failure, because we had monitoring outside the Akamai we knew this was an external problem:
And we knew when it was fixed:
Here is an example of a RUM monitor which reflects a spike in pageload time that was reported from customers browsers which accurately alerted us to a user experience problem:
Here is our 7 day Up Time score based on Synthetic monitors:
We would not be able to show this score with confidence without our suite of New Relic synthetic monitors
We also are depending on New Relic alerts as the engine to power our public status page status.redhat.com
These are a few of the many reasons why we love New Relic Synthetics and Browser on the Red Hat Customer Portal.