How to go further
How to go further
This guide has made recommendations based on best practices and commonalities among companies deploying data observability. However, every company’s environment and needs are different, so it’s important to understand how to continue adding observability where you need it most.
When a data incident occurs and it isn’t caught by a monitor, consider adding new monitoring to that table. Use Metaplane’s column-level lineage to identify the columns upstream that create that data, and add monitoring there to catch the issues closer to when they occur. This is a reactive but effective way to ensure you don’t encounter the same problems multiple times.
The more proactive approach is to consult the various data experts in your organization outside of the data and analytics engineering teams: the end users themselves. These end users lack the architecture-wide view of the data that your team has, but they are very familiar with a subset of it. Working backwards from the monitoring needs of these teams is a great way to identify the data points of greatest importance and set up the monitors necessary to ensure its quality.
Next steps
Metaplane can run a workshop for the various teams in your company that use your data to help them understand the power of data observability and get them thinking of the key monitors they would like. Reach out to your Metaplane account manager to set up a demo and workshop to the extended company.
For your existing monitor deployment, continue to “Mark as normal” when an alert is raised on an expected data pattern, add coverage for edge cases missed by your core monitors, and tune the frequency and sensitivity of monitors to match the urgency and scale with which you would respond to an incident. If there’s any functionality not currently available that you would like to see in Metaplane in the future, let us know! Our customers are key drivers of our roadmap. The platform learns and grows with you, and it should always be making your life a little easier.