Workloads in Public Clouds

Posted on Apr 16, 2020 by Amar Tumballi
3 minutes read.

Sriram Subramanian posted some of the key findings from IDC’s Public Cloud Spend research and 2 things stand out

The entire PaaS spend on public cloud infrastructure was on a handful of workloads: Structured Data Management, App Dev & Testing, and Data Analytics (Structured and Unstructured), during the year 2018. App Dev & Testing workloads are also expected to grow on public cloud infrastructure, driven by the adoption of cloud-native development practices, growth in cloud-native applications, and Dev/ Test use cases.

The pace of growth and transformation in the public cloud story has a lot of steam left in it. And as more enterprises undertake the transformation, it is important to enable a solid framework of practices which result in successful outcomes. In conversations with our users and customers we have seen the importance of the concept of "speed of transformation". Or, how quickly can they lift and shift their application and workloads into the public cloud while taking advantage of established body of knowledge in SRE/Ops manuals.

Application development and testing methodologies create high velocity and high churn workloads. Especially when integrated with event driven flows they integrate declarative deployment methods to create scenario based tests and failure testing. Designing for easy configuration and deployment of infrastructure is intimately tied to adoption of well known design patterns. At Kadalu.IO our focus has been delivering the most simple and complete experience for provisioning storage on Kubernetes. We bring our knowledge of containerized Gluster for large and scaled workflows and ensure fault tolerance and recovery mechanisms are baked into the experience. What we do expect is that enabling a predictable experience at the infrastructure layer allows us to engage in design decisions around the application development layer.

The best set of infrastructure to complement application development and testing routines have to be built around all aspects of the software development lifecycle. We strongly believe that we cannot look at our project and product in isolation. We have a choice - focus on delivering the best experience around storage in kubernetes. Or, engage in partnerships with other members in our ecosystem to deliver the best integrated experience. One aspect to that is being able to make changes that deliver an opinionated experience for storage. For instance, the work we are putting together with an external control plane (see RFC here) for stand-alone Gluster deployments is aimed at making available the same class of features which we include for natively containerized deployments. The similarity in ease of use and deployment reduces the complexity of training and operations that Ops team members have to invest in and maintain over a period of time.

The "public cloud" conversation is no longer an "if". It is about "how soon" and the playbooks which enable technology teams to make rapid deployment possible. In the Kadalu.IO community we are engaged in building up more connections so that our users can publicly share benchmarks and performance tuning data which can be drawn into enterprise deployments. Join the conversation today!

Join the Kadalu team as we plan, design and develop new features. We are available on our Slack instance,

Amar Tumballi

GlusterFS and Kadalu Maintainer

Website| Twitter| Github

© 2022 Kadalu Community( All Rights Reserved.