Zero Downtime Architecture That Scales
Okta must be available for any other app to be accessed and therefore there’s no good time to be down. As a result we are built for high availability and scale and deliver a 99.9% guaranteed uptime. When we say 99.9% we mean 99.9%. No ifs, ands or buts. With Okta there is zero planned downtime. The Okta service never shuts down for maintenance purposes.
The Okta service never shuts down for maintenance purposes.
Okta’s Zero Downtime Architecture is 100% multitenant, stateless, and extremely redundant.
Learn More ›
A Measure of True Availability — No Maintenance Windows
At Okta we realize that our customers rely on the user authentication component of our service to access their business critical applications. With that in mind, we designed the Okta service so that it never shuts down for maintenance purposes. The Okta service and our underlying cloud services platform is designed to achieve zero planned downtime while also supporting the introduction of new functionality on a regular basis.
As a result when we track and publish availability, it is really the true availability of the system. Our calculations are not adjusted to account for planned downtime — because at Okta there is no such notion.
How We Do It
There’s no magic wand for building a highly scalable, highly available service. Every system tier must be built and tested to operate with these goals in mind.
Okta’s app tier is stateless and load balanced. Within each availability zone, at least three app servers operate interchangeably, and more can be added when load demands.
Okta uses Amazon Web Services as its hosting provider. AWS provides all the facilities needed to build a highly available, scalable service, including rapid server image instantiation, multiple availability zones, multiple network providers, and multiple physical locations. Okta has built its service to fully utilize all of these capabilities.
Each database has been built and tuned for high availability. They are all run in redundant threes within a single availability zone, with one active at any given moment. Additionally, databases are actively synced between availability zones for global failover.
It is ultimately people who make a service run. Okta’s operations team is one of the best in the business with deep experience in running some of the most trusted cloud services in the industry.
As part of our commitment to security, Okta has used the SOC 2, Type I and Type II processes — formerly known as SAS 70 — to successfully audit the operational and security processes of our service and our company. The detailed results of this stringent process are available upon request — simply email firstname.lastname@example.org.
Okta truly values transparency and building trust based relationships with all of our customers. This page is just one piece of that relationship and we would very much like to continue that conversation.
What else you would you like to see here? Send us an email at email@example.com today.