Explaining Time Series Data
The aim of these notes is to share a bit of the podcast conversation — SE Radio 623: Michael J. Freedman on TimescaleDB — focused on the explanation of the time series data. I work with that data daily in my support work, looking into the monitoring dashboards of the cloud services. It’s good to have it explained.
Information collected over time that’s changing and you want to look at the history.
- TimeSeries are defined by use case and priority, by how important it traditionally is — for a company, and how you use it.
- One of the big places where the notion of a time series emerged in the last 15 years and where you see some of the specializations is in server initially started in what people called APM.
- And that became broadened into observability
- There are both companies that started with AppDynamics and Dynatrace and New Relic and more recently Datadog.
- And you often also see open-source software, probably the best known is Prometheus and now it’s being standardized, and Open Telemetry which is focused on collecting the data.
- The use case of this is typically an internal one
- It is important that you have observability, but if there are any blips in your observability architecture, it’s not great because you’re flying blind.
- But your customers are not offline. It's not like your SAP is down. Or you are getting BSODs across the enterprise (Hello Crowdstrike)
And so the use case is this vertical solution where the end user might be the SRE or the DevOps team. What they really care about is whether there are dashboards or alerting system in place.
And so when some people think about time series data, they immediately go to the observability use case, particularly if they’re an engineer who deals at all with operations.
> On TimescaleDB: this is what TimescaleDB is not in that we saw this use case of throwing your data somewhere, building some dashboards on this metric, not as mission critical, no need for as much flexibility, and no need to often integrate this with other important business data as a very different use case than what Timescale started at and continues to see.