November 2023 Summaries

The article discusses building metrics with TimescaleDB to understand system performance better. It highlights common user concerns such as function status, delays, throttling, and misconfigurations. The author explains the choice of TimescaleDB over other tools like Prometheus and InfluxDB due to its compatibility with Postgres, existing feature usage, and local development solution. The requirements for the MVP release include function throughput, SDK throughput, and throttle indicators. The author also explains the choice of storing data in a flat, schema-agnostic structure over other methods like keeping each entry as a record or using Counter, Gauges, Histograms. The article further discusses challenges faced during implementation, such as Timescale's continuous aggregate feature and high cardinality issues. It also touches upon the need to remove tags from metrics due to changes in context. The author concludes by sharing their positive experience with TimescaleDB and its cloud offering.

This article discusses error handling in programming languages, comparing thrown errors with errors as values. It highlights that while Python traditionally uses thrown errors, returning errors as values can be more effective for handling and documenting errors. The author demonstrates how to handle errors as values in Python using different approaches such as tuples, the Result library, and unions. They conclude by explaining why they chose to use unions for error handling in their Python SDK.