Snapshots provide a shareable comprehensive view into system behavior for a given point in time. They’re useful for capturing and communicating system information to facilitate cross-team collaboration and help understand system behavior, both in real-time and historically.
Every Query made in Explorer automatically saves the results in what is called a Snapshot. A Snapshot is persisted so that it can be shared and revisited at any time. The Snapshot includes all the information about aggregate system behavior that you see in Explorer: the set of traces, the latency histogram, and the other analysis provided in Explorer.
Snapshots are intended to help you and your teammates quickly communicate with each other about system behaviors. Sharing functionality is built directly into to Snapshots:
URLs to particular Snapshot are valid for as long as the Snapshot is persisted. Explorer maintains a list of your most recent Snapshots:
Snapshots are created every time a Query [link] is run in Explorer. The results you see are sharable for the duration of your project’s data retention period.
Snapshots collect enough information from the running system to provide a aggregate picture of system behavior. Precisely how long this collection process takes will vary from system to system, but in normal circumstances will be on the order of seconds. See the Query documentation for more details.
During an incident, Snapshots will capture any and all queries made in Explorer. If in your incident playbooks you keep a link a to relevant query for the incident, just visiting that link and letting Explorer run the query will automatically create a highly detailed view of the system behavior at that moment. These Snapshots can be shared in real-time as you pull in teammates and share the clues and theories you have.
After an incident has been resolved, Snapshots can be linked to in post-mortems to revisit exactly what happened at the time. This changes the process of reconstructing the data of what happened from the recollections of the individuals operating in a high-stress environment to revisiting the actual system behavior at the time. This helps in the cases where incident mitigation and incident resolution diverge. Snapshot links recorded in post-mortems allow a way to rewind back to see what was really going on at the time of the incident.
Snapshots capture the full results from any Query in Explorer. This allows you to both examine the detailed span, log, and payload information of each individual end-to-end transaction as well as view the aggregate system behavior. This highly detailed view of production behavior, across many example transactions, can help you make informed decisions about code and service changes that otherwise would be difficult to infer from the code itself.
Snapshot functionality requires Satellite release 2018-07-03_21-59-17Z or later.
Please see the LightStep Public APIs page for details.