Persistent Storage for Containers

According to Gartner Group, by 2022, more than 75% of global organizations will be running containerized applications in production.  To maximize the potential for containerized applications, under any orchestration tool (Kubernetes, Docker Engine, etc.) the underlying storage systems must address three fundamental requirements:

  • Persistence
  • Portability
  • Performance


Persistent Storage for Containers

Containers were traditionally designed as ephemeral resources for development and test. As container benefits like the separation of infrastructure from application become obvious, there is a natural desire to move from experiment to production. This requires a definition of a specific container as persistent.  Using functions inherent to the Pavilion system, it is possible to define Persistent Volumes with Persistent Volume Claims that have specific Service Level Agreements for storage read/write bandwidth, IOPS, volume size and non-disruptive volume expansion with thin provisioning.  This allows for applications to be stateful, manageable and scalable.


Portability for Containers

Zero-footprint snapshots are particularly valuable for running containerized applications.  Administrators can orchestrate movement of snaps and clones to any host across any standard Ethernet or InfiniBand network.  These clones are not restricted to the pod where the master volume is being used.  This same snapshot process allows for isolation of container backups at a volume, pod or cluster level.  Persistent volume containers can now be portable to any team in the organization without concern for host location or storage capacity consumption on the array.


High Performance Storage for Containers

Storage Area Network (SAN) technology has proven to be the most widely accepted approach to persistent storage.  But SANs, especially fibre channel-based implementations, carry the baggage of support for legacy hard disk protocols like SATA and SAS which add significant latency to containerized applications.  NVMe-oF is a very low latency disaggregation alternative where technologies like Pavilion Data’s HyperParallel Flash Array shine with high-bandwidth, low latency storage for containers.  A Pavilion system can deliver 120 GB/s of read and 90 GB/s write bandwidth, with 20 million 4K Read IOPS at 40 microseconds of latency.  This means that production containerized applications can scale unbounded by the underlying persistent storage infrastructure.

As you consider deployment of production containers, keep in mind the importance of the underlying storage and implications for operating at scale.  Here, Pavilion Data has your back.  To learn more about persistent storage for containers, please refer to our latest White Paper.