Portworx provides synchronous replication of data across multiple nodes in a cluster for high availability. It can be used for stateful applications and databases needing persistent storage. Big data hardware often includes lots of local storage that expands with the number of nodes, but new container storage interfaces will provide a vendor-neutral way to connect storage to orchestrators like DC/OS and Kubernetes. When deploying big data on DC/OS, choices of storage include only local storage or a hyperconverged option using Portworx to provide a single pool of storage. Best practices include using health checks, operators/frameworks, measuring performance of local versus shared storage, and considering encryption.