Sometimes scale get contrived as huge or some quantity of capacity in IT that few shops will ever get to. I think big or small all companies need to the ability to scale. The ability to scale allows customers to buy want they need, when they need it and most importantly use it right away. It can be 6 TB or 60 PB, it’s all relative.
The prized gem for Nutanix and it’s ability to scale revolves around Apache Cassandra(NoSQL) and paxos. Nutanix stores it’s metadata in Apache Cassandra There is a good write up on how Paxos works with NoSQL on Nutanix on the Nutanix Bible. I really enjoyed the ending of a recent article, “Next gen NoSQL: The demise of eventual consistency?”
The next generation of commercial distributed databases with strong consistency won’t be as easy to build, but they will be much more powerful than their predecessors. Like the first generation, they will have true shared-nothing distributed architectures, fault tolerance and scalability.
Why did I enjoy it? Because this is what Apache Cassandra (NoSQL) and paxos is giving to Nutanix today. NoSQL is powerful tool for responding to change and combined with paxos all worries go away. NoSQL’s ability not to need a strict schema allows Nutanix respond to change very efficiently in terms of:
Failures – Nutanix Cassandra has self ring healing in 3.5 where the metadata is evenly distributed. If cassandra process on a node is down for more than 5 minutes. Medusa will trigger the process of detaching the node from cassandra ring. Once the node is detached from the ring, we are ready to take another failure and still remain available.
Upgrades – Only that is constant is change! Nutanix is rapidly adding features and our customers can’t afford downtime. I just read a couple of days ago of company adding dedupe to their product line and the upgrade needed planned downtime. NoSQL allowed Nutanix to add SHA1 hashes to the metadata and carry on to provide inline dedupe without downtime.
Scaling – Nutanix can scale compute and storage at different rates with a variety of different nodes. The process is the same for all the nodes. Hit the expand cluster button, enter in three IP’s, add to the compute node to vCenter. You also have the ability to automate the whole thing! Keep in mind this process is the same for ESXi, Hyper-V and KVM.
Scaling is the ability to respond to business change.