Dynamic PostgreSQL is a managed PostgreSQL offering on Timescale that is built for your production workloads. Dynamic PostgreSQL is 100% PostgreSQL with no alterations to the core database. It comes with Timescale's dynamic compute and usage-based storage, giving you the scalability of a serverless database with the performance and cost-effectiveness of a statically allocated DBaaS.
Dynamic compute means you choose a compute range with a minimum and maximum instead of a single static compute option. Your database always has the minimum compute allocated to it. As your load increases and your application demands more of your database, your database can use up to the maximum of your range with zero delay.
With dynamic compute, your database always has the effective memory corresponding to the maximum of the compute range. For example, in a 4 - 8 CPU configuration, your database has 32 GB of effective memory.
In this model, you only pay for your minimum compute, and are metered for any usage above your minimum.
Dynamic PostgreSQL is currently using early access pricing, where any metered usage above the minimum is free.
Dynamic PostgreSQL comes with Timescale's usage-based storage. With usage-based storage, you only pay for the storage saved on disk, rather than the amount of disk space allocated. Usage-based storage is billed hourly per gigabyte of data.
Dynamic PostgreSQL is designed for continuous production workloads. There are four main types of database workloads: uniform, variable, bursty, and intermittent.
Uniform workloads have a constant CPU usage. Dynamic PostgreSQL serves these workloads efficiently if the usage is provisioned within the compute range. This allows you to only pay for the compute you use, rather than overprovisioning.
Variable and bursty workloads have some kind of seasonality or spikiness. Dynamic PostgreSQL is great for these workloads as it allows you to scale seamlessly and only pay for the usage fluctuation above your minimum.
Intermittent workloads are workloads that happen only occasionally, don't need in-memory caching, and can tolerate a cold start. Serverless databases are generally a better fit for these workloads.
With Dynamic PostgreSQL, your bill has 2 components:
- Your storage costs, billed in GB/hours
- Your compute costs, which is your base compute plus any fractional CPU usage above your base, up to your max, in CPU/hours
Found an issue on this page?Report an issue!