On MariaDB Kubernetes Operator. Q&A with Saravana Krishnamurthy
Q1. Why is Kubernetes becoming so important?
With the recent evolution of the DevOps model, companies are now incentivized to adopt modern application development platforms that enable greatly increased productivity and agility. Enter Kubernetes (K8s). Battle-tested as Google’s Borg platform, K8s has proved its mettle and successfully supported thousands of DevOps engineering projects on millions of server instances. Once Google open-sourced K8s, it spiked a huge interest in the cloud and open source community. In addition to community involvement, Google also continues to heavily invest in the growth of K8s. Now, the PaaS has become a defacto standard for running stateless and stateful applications and microservices in demanding enterprise environments. Its influence cannot be understated; even some of the earlier container platforms such as Mesosphere, Docker swarm and PCF support K8s to retain relevance in the container world. At MariaDB, we consider K8s to be a bedrock technology that enables enterprises to realize their multi-cloud dreams.
Q2. What are the main benefits for customers in using Kubernetes?
K8s provides many benefits for our customers, but as a database company, there are two main benefits that customers have found especially valuable:: multi-cloud portability and stateful application support.
With K8s, we’re able to offer a native, high-performance, first-class database experience to all of our customers, and enable a multi-cloud approach. Customers can start with on-prem K8s deployments and migrate their applications to the public cloud of their choice over time. Alternatively, some customers prefer to run DevOps, development, testing and staging in public cloud K8s platforms and move production on-prem for enhanced security and governance. But as the complexity of these applications running on-prem grow, customers often consider a multi-cloud approach, be it private or public. In these cases, K8s is a perfect PaaS layer that provides cloud portability thereby enabling cloud elasticity for customers.
Stateful application support is another major boon for customers. Container platforms such as Mesosphere, Docker, and PCF are great for running stateless applications such as web servers, but are less effective at running stateful applications such as databases. The challenge comes from the need for persistent layer access when containers failover to another host or virtual machine. K8s solves this issue with StatefulSets, the workload API object that guarantees the stickiness of a Pod to its persistent layer.
Q3. Does Kubernetes reduce the risk from lock-in?
It does to a large extent. I think of it as a cloud operating system that provides a standard interface for network, storage and security. Applications developed with K8s can run on any of its derivatives (such as OpenShift/Anthos, etc.) with little to no modifications. Additionally, many of the public cloud providers such as AWS, Google and Microsoft provide K8s cluster services, and Google’s recently announced open cloud platform Anthos is indicative of its commitment to standardizing K8s across all cloud platforms.
Q4. How does it (Kubernetes) work with your new MariaDB Kubernetes Operator?
The MariaDB Kubernetes Operator enables our customers to more quickly deploy and manage MariaDB Platform in the cloud, and facilitates complex MariaDB topologies such as single server, master/slave and MariaDB ColumnStore with a single user module and multiple performance modules in a K8s environment. The MariaDB Kubernetes Operator provides a Custom Resource Definition (CRD), extending the K8s API to use MariaDB. It also provides a Docker image that integrates the Helm Chart and CRD into a controller to create, delete and update MariaDB containers in a K8s cluster. It has been tested extensively on GKE, Azure and OpenShift, and it was one of the first operators to achieve RedHat OpenShift Operator Certification.
Q5. Why does achieving Red Hat OpenShift Operator Certification matter?
OpenShift is probably the most widely supported K8s platform currently available to enterprises. We have many joint customers using OpenShift that are eager to deploy MariaDB in OpenShift at a larger scale. Having a certified operator and corresponding container images provides these joint customers a streamlined method to deploy MariaDB in a supportable environment.
Q6. How does it (Kubernetes) fit into your overall cloud strategy?
We see K8s as a key enabler of our multi-cloud strategy. We developed certified container images and operators early-on to provide our customers with enhanced deployment flexibility. Cloud database deployments are growing at a rapid pace, and with K8 we’re offering an unmatched database experience on the cloud of their choice. According to a 451 Research report, it is estimated that by 2021 on-prem bare-metal database deployment will shrink from 78.9% to 27.6%. However, private cloud deployments will grow from 5.3% to 24.5% and public cloud deployments will grow from 5.9% to 47.9%. Customers migrating to the cloud can rest easy; MariaDB is committed to ensuring their success across cloud platforms.
In the last few months, our Fortune 1000 customers have expressed tremendous interest in deploying MariaDB on K8s and OpenShift to help transform their development and production platforms. As a result, application developers, data analysts, data engineers and database administrators are able to self-serve and increase operational agility.
Q7. You are working on MariaDB’s upcoming public cloud service, SkySQL. What is it?
SkySQL is a self provisioning public cloud service that maintains our extremely demanding quality of service expectations and is fully elastic with more autonomous database capability added over time. Thanks to our MariaDB Kubernetes Operator, SkySQL grows with all third party technologies that are part of the K8s and Docker ecosystem—coalescing open source innovation to deliver community-driven cloud performance. SkySQL is frictionless, highly scalable and enables our customers to spin up databases on the fly without having to worry about their database deployments in terms of server and storage characteristics. Instead, they can spend more time focusing on their business.
SkySQL disrupts cloud conventions and will free customers from the constraints imposed by cloud vendors. When customers encounter issues with public cloud platforms today, they experience non-existent—or at best anemic—support from public cloud vendors. It’s understandable—a public cloud is essentially a marketplace where public cloud vendors offer a bag full of tools, many of them open source, without proper support infrastructure behind those open source products. Often, dissatisfied customers will approach us for support and we end up stepping in to solve customer issues. With SkySQL, customers will know their databases are equipped with the incredible quality of service they have come to expect from MariaDB.
To top it all off, customers running MariaDB on public clouds typically end up with older versions and without the latest bug fixes and patches. Many of our customers graduate from a single server to deploying complex topologies such as master/slave, Galera or ColumnStore, and it is impossibly optimistic to deploy these complex production clusters in public cloud environments and expect robust support from cloud vendors.
SkySQL will also take care of production grade activities such as backup/restore, snapshotting, rolling upgrades, timely security patch updates, security/governance and world-class support infrastructure that our customers have come to expect. As a multi-cloud deployment platform that will support multiple public clouds, SkySQL is a game changer in the public cloud RDBMS market.
Saravana Krishnamurthy is the VP of Product Management for Cloud and Analytics at MariaDB Corporation. He has more than 20 years of experience in the database, Big Data and infrastructure software industry. Prior to MariaDB, he held leadership positions at various companies such as BlueData (acquired by HPE), Xplain.io (acquired by Cloudera), Pivotal/Greenplum, Motorola and Sun Microsystems.