Robust Applications in Mesos Using External Storage

Explore the intricacies of using external storage for robust applications in Mesos in this informative conference talk. Delve into the challenges and benefits of utilizing direct attached or external storage for persistent applications like SQL and NoSQL databases in container schedulers. Examine the impact of storage type choices on production environments, from initial deployment to day-to-day operations and system failure tolerance. Learn about container advantages for stateful applications, including consistent environments, health monitoring, and automated rollouts. Compare initial deployment and day-to-day operations using local disk versus external storage, focusing on data locality and hardware failure scenarios. Investigate the specific challenges faced by NoSQL and key-value stores, including scale-out and scale-in operations, monitoring, automated recovery, and bootstrap processes. Understand the potential pitfalls of repair processes in applications like Cassandra, and discover strategies to minimize vulnerability windows during node failures, disk issues, compute failures, network partitions, and maintenance.

Intro
Container Advantages Make Sense for States Too Container attributes Orchestration can adt . Consistent environment - Health mentoring same anywhere • Automated rolous and Dependency management-packaging. Declarative configuration
Initial Deploy Using Local Disk . Simple and straightforward • Performance based on compute node storage capabilities • Targeted deploy based on resources
Day 2 Operations Using Local Disk Data locality!
Day 2 Operations Using External Storage External volume moves with the Container - Hardware failure
What about NoSQL & KeyValue Stores? Initial Deploy - Local disk: Same External storage Same - Day 2 Operations?
Making operational aspects easier Scale out & Scale in - Monitoring - Automated recovery - Bootstrap and rebuild • Elephant in the room!
repair process is expensive - Your application... Slows down Grinds to a halt Can even bring down Cassandra If you aren't careful
Cassandra node failure Bad Disk Compute failure Network partition Maintenance
Minimize window of vulnerability • Run node repair tool Not a full node rebuild Delta - Migration time