Built at UC Berkeley, Mesos was designed for operations at very large scale. It was tested and refined under the crushing loads of Twitter and Airbnb, and now supports some of the largest applications in the world.
Mesos supports different types of workloads to build a truly modern application. These distributed workloads include container orchestration (Mesos containers, Docker, and Kubernetes), analytics (Spark), big data technologies (Kafka, Cassandra) and much more.
Mesos’ two-level architecture allows it to run existing and new distributed technologies on the same platform. The two-level architecture allows organizations to build their own operational logic in their apps, dramatically simplifying operations.
Mesos began as a research project at UC Berkeley, the birthplace of BSD Unix. The project was started at AMPLab by then PhD students Benjamin Hindman, Andy Konwinski, and Matei Zaharia, with professor Ion Stoica. Inspired by Google’s proprietary Borg system, the project’s goal was to create an open-source kernel that simplifies building and running distributed applications at a very large scale and treats the entire datacenter as a single giant super computer, while also maintaining an extensible architecture.
A key design criteria of Apache Mesos is its two-level scheduler architecture, making it easier to operate, scale and extend.
Traditional monolithic schedulers maintain the complete state of the application and infrastructure underneath, while also performing workload placement logic. This architecture makes it very challenging to scale and even harder to introduce new features and capabilities.
With a dual-level architecture, Mesos handles low level infrastructure scheduling operations, while another layer on top (The framework) handles all the application specific operations and logic. This architecture has multiple benefits:
Mesos can support different types of distributed workloads such as container orchestration (Mesos and Docker containers), analytics (Spark), stateful big-data technologies (Kafka, Cassandra) and even different types of operating systems (Windows and Linux), allowing organizations to simplify operations and achieve higher resource utilization.
By having Mesos handle low-level infrastructure logic and delegating the application-specific logic to the framework, Mesos can easily scale to tens of thousands of nodes, which is why it has been reliably powering some of the largest web-scale platforms in the last couple of years.
As new distributed technologies are introduced to the market, organizations can easily introduce and adopt new technologies as applications or frameworks (such as Kubernetes), and include any application-specific scheduling and operational logic, making Mesos a true future-proof platform for distributed technologies.
Apache Mesos is used in many large scale production deployments
Easily roll out new datacenter services, from our strong and growing ecosystem.
Complement Mesos with enterprise capabilities such as security, monitoring, user interface, etc.
Easily run and scale with a production proven distributed systems kernel.
Run DC/OS on any Infrastructure, and standardize your operational and application development experience everywhere.
Validated Architecture
Use a validated, tested solution with best practices, and reduce snowflakes in your infrastructure.
Quick Production
Stand up and scale a complete solution and get to production in days instead of weeks.
Application Packaging
Deploy, operate and upgrade complex technologies such as Spark, Kafka and Cassandra with a single command.
Simplified Management
Use the DC/OS CLI/UI/API to easily monitor and operate your entire infrastructure.
24/7 Support
Get top-notch support from Mesosphere & core committers, and influence the roadmap.
No Lock-in
Run all Mesos frameworks on DC/OS, & ensure no vendor lock-in with our open-core model.