• Technical
100% Availability or Your Money Back with Instaclustr

Instaclustr now offers availability SLAs of up to 100% with associated service credits of up to 100% of monthly fees.

We have recently reviewed our SLAs and decided that our previous SLA levels did not reflect the performance that we expect of ourselves or that our customers need for the mission-critical usage of Cassandra. We are now offering a tiered series of SLAs based on the size of the cluster that you have under Instaclustr management.

For clusters with 12 or more nodes, we now offer a 100% availability SLA with up to 100% service credits for any month where we fail to meet this promise. This is an industry-leading offering which reflects:

  • our confidence in Cassandra as a rock-solid platform for mission critical applications;
  • our understanding that for many customers the choice of Cassandra and Instaclustr is a “bet the company” proposition and our desire to respect the confidence you place in us; and
  • our confidence, based on our 20 million+ node-hours of experience running Cassandra, in the management systems that Instaclustr has in place to ensure availability of your Cassandra service.

This tiered SLA model is only possible because of Cassandra’s inherent scalability, reliability and manageability and the fact that, unlike most other database systems, Cassandra is actually more manageable and reliable as the cluster scales out.

Of course, being an SLA, there is some fine print to consider but generally, these conditions reflect good practice in developing and managing a Cassandra based-application for the highest levels of availability. Some specific conditions to be aware of for the 100% availability SLA are:

  • your keyspaces must be configured with a replication factor of at least 5;
  • availability is based on successful reads/writes with QUOROM consistency; and
  • you need to work with us to manage capacity usage on your cluster to maintain sufficient free disk space for Cassandra operations (generally <70% used during normal operations).

Should these conditions be more than you feel are necessary for your requirements, then our 99.95% availability SLA remains but is now covers QUOROM consistency operations rather than consistency level 1 (for clusters with 6 or more nodes and a replication factor of 3).

We have also introduced latency-based SLAs and upped the expectations on the level of support we provide to most customers. More on that in future posts and the full SLAs are available here.