Skip to content
Kubernetes Storage

Don’t treat Kubernetes storage as an afterthought: Turn to persistent container storage for high performance and availability

Founder and CEO of StorageOS

In their quests to harvest the business benefits of hybrid and multicloud services, enterprise cloud teams face significant challenges in managing the orchestration of containerized applications and resource provisioning in a performant and cost-effective way.

Organizations turn to Kubernetes to accelerate development, test, and production cycles. Storage is often an afterthought, and that’s a problem. When storage is separate from Kubernetes clusters, cloud applications may experience excessive latency with subpar performance for critical applications. That’s a non-starter for financial services. An application in a Kubernetes cluster, such as an enterprise database, may experience slow input/output (I/O) or throughput because of the architectural imbalance caused by external storage.

In the cloud-native world of Kubernetes, storage isn’t something you attach to servers or hypervisors. It works best when it is application centric. Storage needs to be portable to reduce infrastructure costs and support automated provisioning.

How do we get there? Containerized applications previously had limited options for persistent storage capable of supporting high-performance enterprise applications that require high availability, automation, and rapid failover.

Today, many organizations embrace the benefits of software-defined storage (SDS), which in a Kubernetes environment, allows nodes to scale up or down as needed — with the rate of change and speed that enterprises require. But, without cloud-native SDS built for containerized applications, organizations must continue to invest in infrastructure, and developers remain unable to self-provision storage. That can reduce their agility in responding to changing business opportunities.

How to evaluate containerized storage solutions?

A containerized, cloud-native storage solution must excel at the following key performance indicators (KPIs) and attributes:

  • Performance. Chris Evans of Architecting IT conducted a performance benchmarking test of cloud-native storage solutions for Kubernetes. Evans looked for a solution in the trials that could deliver “greater throughput, bandwidth, and with lower I/O latency.”

  • Cost savings. Enterprises looking to eliminate their storage capital expenditures (CapEx) can routinely save tens to hundreds of thousands of dollars annually by switching to cloud-based commodity storage rather than continuing to purchase their server or network-attached storage.

  • Scalable. Using disaggregated consensus, each volume should independently manage placement and failover decisions. Operations must happen in parallel, which builds support for large clusters. Failover must occur in milliseconds, supporting the high rate of change demanded by cloud-native and OpenShift environments.

  • Secure. Ops teams need end-to-end encryption of data in flight and at rest using only known keys to the application. That saves organizations the step of handing them over to a service provider. Ideally, this happens in a highly optimized, low latency data plane.

  • Reliable. Downtime should be significantly reduced with replicas, allowing for non-disruptive recoveries from failures and applications to achieve durability and consistency.

  • Native Kubernetes integration. Look for a solution that delivers declarative self-service storage, allowing enterprises to build anything “as a service.” This generates reproducible deployments that adhere to cloud-native design patterns and best practices.

Built to scale

SDS is popular and proven technology, ideal for cloud-native environments. According to a Technavio report, SDS is a transformational development with a 27% compound annual growth rate because it simplifies storage management and drives “a reduction in costs, enhanced scalability, and higher efficiency.”

We built StorageOS to capitalize on these benefits. When we launched StorageOS six years ago, we sought to fill the architectural gaps preventing enterprises from shifting high-performance applications to the cloud. We not only removed those barriers, but we also outperformed our leading competitors across all three performance tests conducted by Architecting IT’s Evans.

We provide persistent storage to applications in containerized environments, a capability that enables DevOps or CloudOps teams to create and manage highly performant and reliable cloud-native applications. Our customers embrace Kubernetes to take advantage of consumption-based pricing models and to use self-service and automation, providing them greater control over their workload costs and application performance. For example:

  • No latency, no problem. How do you establish the reliability and integrity that a high-volume trading platform requires when you cannot lose a single transaction in a stream of millions of daily trades? StorageOS provides deterministic low latency to meet trading platform KPIs for reliability and performance. As a result, one customer’s trading platform achieved a 75% reduction in latency for its message queues.

  • We work with several cloud-native service providers who manage Kubernetes platforms, such as OpenShift, for their customers, providing the storage engine with enterprise-class encryption and security. A customer achieved 20x faster performance running StorageOS than it experienced with its existing storage system.

How StorageOS works

How it works

Each StorageOS container includes an intelligent control plane that can make dynamic changes and a data plane that provides replication, compression, and encryption for stateful applications.

In StorageOS, we’re building out the vision of a platform-agnostic storage engine that can provide enterprise-class storage features that you would expect from a traditional storage system—except now; you can run it anywhere.

Customers tell us that managing the velocity of change is a constant challenge in dynamic continuous integration and delivery (CI/CD) environments. Enterprise workloads scale up and down and require frequent adjustments and immediate action in functions such as store, create, delete, and move.

We meet these challenges with a cloud-native control plane that applies automation to volume provisioning and cluster operations, enabling these actions in milliseconds across large clusters.

StorageOS aggregates storage across all nodes in a cluster into a pool, allowing volumes to be provisioned from the pool and containers to mount those volumes from anywhere. Our storage engine transparently redirects reads and writes to the appropriate volume, so the container is unaware of accessing local storage or remote storage. Volumes are also thin provisioned to avoid consuming disk space unnecessarily.

Sustaining a highly available environment is nearly impossible without automation. Under the hood, StorageOS replicates data simultaneously across different nodes. With StorageOS, if a database pod fails, a cluster can recover from its degraded state as quickly as Kubernetes can restart it, ensuring that the pod data stays available.

Transform your operations

Avoiding vendor lock-in is one of the key reasons enterprises embrace Kubernetes environments like OpenShift. That’s just the starting point. With StorageOS, you can add to those benefits and scale-out a multicloud or hybrid cloud strategy that delivers the self-service operational advantages that were your original objective.

StorageOS combined with OpenShift provides enterprises with an automated, end-to-end infrastructure platform that provides much more granular control over their storage provisioning and spending. We were one of the first companies to achieve Red Hat certification for the StorageOS operator. Now, a robust, secure, and performant cloud-native enterprise storage platform is available to you with a single click.

We are a leading software-defined storage solution deployed via container and can run anywhere without dependencies on proprietary kernels or hardware. We are environmentally agnostic, unlike other cloud alternatives you may encounter. This means regardless of the environment, developers enjoy the same software-defined storage experience and capabilities that deliver unmatched performance and value for enterprise applications.

To learn more or evaluate a free trial of StorageOS, please visit our product page on Red Hat Marketplace.

Article originally featured on Red Hat Marketplace