On MariaDB SkySQL. Q&A with Sameer Tiwari

Q1. You have recently announced the availability of MariaDB SkySQL, a database-as-a-service (DBaaS). Is it a Cloud service? How does it relate to the MariaDB Platform?

Sameer Tiwari: MariaDB SkySQL is a fully managed cloud database that brings the full power of MariaDB Platform to the cloud. SkySQL supports transactions, analytics and hybrid transactions/analytics all at any scale and in one cloud service. It’s based on MariaDB Enterprise Server, the core component of MariaDB Platform, to support mission critical workloads in the cloud. Customers get access to the latest versions of MariaDB so they always have the latest bug fixes and security patches. Plus, SkySQL comes directly from the source, the same team building the database and team that’s trusted to support the largest enterprises in the world. There’s nothing else like it on the market.

Q2. What makes SkySQL different from other Database-as-a-Service (DBaaS) offerings in the market?

Sameer Tiwari: A number of years ago, we discovered that the templates offered in other DBaaS offerings were inadequate. They lacked the richness of MariaDB. That’s when we decided to build our own DBaaS to deliver a quality manifestation of MariaDB in the cloud. One that only we can do since we also develop the database. At the same time, we had Kubernetes rise up as a modern deployment mechanism that gave us a new way to approach building a DBaaS and achieve multi-cloud. These came together in a unique way and the next thing we knew, we had something really strong.

Rather than be hamstrung in the cloud, SkySQL gives our customers the full range of motion they are used to with MariaDB Platform on prem but in the cloud and fully managed by a team that knows the database better than anyone. 

Q3. Can you explain the architecture of SkySQL?

Sameer Tiwari: SkySQL implements a groundbreaking, state-of-the-art architecture based on Kubernetes and ServiceNow, and with a strong emphasis on cloud security – using compartmentalization and indirect access to secure and protect customer databases.

The most important design decision we made was to use Kubernetes as a cloud-agnostic platform for deploying databases. When we look at public clouds, we don’t see infrastructure-as-a-service providers. We see Kubernetes-as-a-service. Amazon has its Elastic Kubernetes Service (EKS). Microsoft has its Azure Kubernetes Service (AKS), and Google has its Google Kubernetes Engine (GKE). Then there is Red Hat OpenShift. Kubernetes is the key that unlocks all of these doors for SkySQL.

The other pivotal architecture decision was to build the SkySQL portal on ServiceNow. There are two reasons why. First, inventory, configuration and workflow management are critical capabilities for a DBaaS, and ServiceNow is the leader in enterprise IT service management. It’s what they do, and they do it best. Second, in order to be truly cloud independent, SkySQL must not be dependent on Amazon Web Services, Microsoft Azure or Google Cloud Platform. By running on ServiceNow, SkySQL sits above the clouds.

SkySQL also runs on a rich security paradigm. It not only relies on cloud native security features, but goes above and beyond to log all access, audit trails, custom umbrella execution policies for binaries in production, and a locked down host from which all access is throttled and controlled.

To dive deeper into the architecture of SkySQL, check out a recent talk I gave on the topic.

Q4. How are you leveraging Kubernetes in SkySQL?

Sameer Tiwari: Early on, we chose Kubernetes as the orchestration engine for SkySQL. This was over two years ago. At that point, many companies were still wondering whether Kubernetes was the right orchestration model for stateful services like a database. For us, we saw the potential for MariaDB as a modern deployment method to be cloud-agnostic. 

It certainly wasn’t easy. We partnered with Google Cloud to launch SkySQL initially on GCP. It turns out we were one of the very few Google customers trying to put system level software on Kubernetes. That partnership allowed our team to work closely with Google Cloud and GKE engineers to optimize SkySQL to run on GCP. 

Kubernetes has been great for providing self-healing capabilities, auto upgrades, high availability, elasticity, predictable deployments, higher resource utilization, a consistent environment across Dev/Test/Staging and Production environments and, most importantly, the ability to move across clouds with very little rewrite of the code base.

Q5. You say that SkySQL is engineered for multi-cloud. How do you accomplish that?

Sameer Tiwari: Simplistically speaking, there are two distinct parts of building a cloud-agnostic and cloud-portable architecture. At the lower level there is a paving layer that is cloud-specific. This includes instance types, cloud specific commands, networking setup, account and billing management, etc. On top of that layer is the cloud-agnostic and portable layer where various products can be hosted without worrying about the underlying substrate.

SkySQL has been designed from the ground up to have this clear demarcation. It has a top layer based on Kubernetes where all the intelligence and “brains” specific to working with each of the MariaDB products are handled by the SkySQL Kubernetes Operator. This top layer can easily be bolted onto new cloud-specific pavements, giving SkySQL the ability to move across clouds with ease.

Q6. Explain SkySQL Power. How are you able to offer DBaaS customization?

Sameer Tiwari: SkySQL Power reverses the paradigm of making a general purpose DBaaS for the masses to that of a custom-built system that caters to a specific user’s requirements. This is best described with a few examples that customers have asked for. A power customer might ask for custom instance sizes to better control costs, or want to brand the UI, create templates for spinning up databases that teams can access, have custom backup/restore policies, request a non-standard topology, or need access to custom configuration parameters.

SkySQL has been designed in a way to be able to support custom deployments and let the customers play with “lego blocks” to build a custom-tailored system to suit their unique needs.

Q7. Who is already using MariaDB SkySQL and for what kind of applications?

Sameer Tiwari: Available since March 2020, SkySQL has been used by customers in 38 countries around the world for a wide variety of use cases including for extra capacity to handle peak workloads, micro-lending transactions combined with analytics for real-time financial analysis, lift and shift from MariaDB and MySQL on premises to the cloud, data redundancy for disaster recovery strategies and migration from AWS Redshift to MariaDB for analytics in the cloud.

Q8. Anything else you wish to add?

Sameer Tiwari: I encourage everyone to give SkySQL a whirl. We are offering a credit to get started on SkySQL today and it literally takes seconds to spin up a new database service. Plus we have tons of materials to get you going, including sample applications you can use with SkySQL. 


Sameer Tiwari is CTO of infrastructure at MariaDB Corporation and leads the development of SkySQL, MariaDB’s database-as-a-service (DBaaS) platform. Prior to MariaDB, Sameer was at Salesforce, where he worked on modernizing the Salesforce cloud architecture to move thousands of on-prem machines to the cloud. He also architected a distributed, scale-out platform for high-speed transactional workloads to replace a proprietary database. In earlier roles, Sameer served as a storage architect at Pivotal and a platform architect at Yahoo.

Sponsored by MariaDB Corporation.

You may also like...