Presentation

Exoscale DBaaS is a database as a service product with a set of data services that can be triggered on demand.

Features

It comes with:

  • Managed data engine with high SLA
  • Selection of plans with upscale and downscale possibilities as well as redundancy options
  • Automatic backup policies
  • Lifecycle management

Terminology

  • DBaaS: database as a service.
  • Service: a service is deployment of a DBaaS flavor.
  • Plan: a plan is a combination of compute resources based on Exoscale Compute instances types and of data infrastructure features (replicas, backup frequency, …).
  • Fork: a fork is the ability to create a new DBaaS service from an existing running service. It enables uses cases for upgrades, data protection, disaster recovery,… scenarios.

Architecture

Database services are launched on dedicated instances but not directly accessible to the end user or visible in your Compute instance list. The database services are therefore not shared and run on single to several machines specifically deployed for your organization depending on the plan.

Service availability

4 services are currently available. These services along with their short names are:

  • Managed PostgreSQL: pg
  • Managed Apache Kafka: kafka
  • Managed Redis™: redis
  • Managed Mysql: mysql

Geographical availability

The Exoscale DBaaS product described in the following guide is available in all current Exoscale zones.

Service level agreement

All plans come with SLA regarding the availability of the data service:

  • all Hobbyist and Startup plans: 99.95%
  • Business and upwards: 99.99%

Known limitations of DBaaS

A number of features are currently unavailable in DBaaS and need to be taken into account while working with it.

CLI/API support and partial portal support

Interaction with DBaaS currently happens with the exo CLI utility and API for the full set of features. The Web Portal provides a partial coverage of the API and CLI possibilities and will progressively catch up with these for Q2 2022.

Service integration

Integrating services with one another is not currently possible.

Multi-zone setup and geo-replication

All services are currently supported on a single zone at once and are not geo-replicated across zones.

Forks are however possible across zones, enabling to start a copy of a service from one zone in another location.

General information about Exoscale DBaaS plans

The plan naming convention indicates the resources of each single or cluster of instances powering your service:

  • Service type: pg, kafka, redis,...
  • Plan: hobbyist-1, business-32, premium-9x-16

Where:

Hobbyist, startup, business, premium keywords differentiate for each service the redundancy and data protection mechanism used. Refer to the product pages to list the available plans and their specifications.

And the optional first digit indicates for distributed systems the number of nodes in the cluster.

Last 1, 2, 4,...32,...,225 indicates the memory associated with each node in the plan in GB. These are indexed on the existing compute instance types available at Exoscale, so a 32GB is a compute general purpose Huge instance.

For example an Apache Kafka service on a premium-9x-16 plan, is a Kafka cluster with 9 nodes each with 16 GB RAM. The other details regarding storage, backup, number of connections, versions and plugins are available in the tables or documentation to be published.

Restrictions and authorizations

Not all plans are available on-demand by default. Some of the plans require an authorization like some of Exoscale other offerings (large instance types, GPU enabled instance,…). You can list the plans that are currently available to your workspace with the command exo dbaas show <SERVICE TYPE> --plans