Our customers are facing two major challenges these days. The most important one is how to enable their business to become truly agile because in the end, this is what makes you ‘win’ as a corporation. It’s the primary thing to focus on but this can be hard: it requires changes not only on the technical side but also on the people and process side. What makes it even harder is that typically 80% of the time is spent on running the operation. This brings us to the second major challenge: how to make the current operation excel through doing what you already did, but in a more and more efficient way.
A container platform can be an answer to this second challenge. As containers are more efficient than the traditional Virtual Machines (VMs), there’s more standardization as managing the Operating System (OS) is centralized, deployment of applications is desired state based with built-in high-availability, and containers are well suited to be used as immutable deployment artifacts in Continuous Delivery (CD) pipelines.
A few years ago choosing a container platform was like placing a risky bet, as there was a handful of competing technologies without a clear sign of which technology would win. These days it’s much more straightforward as the technology ‘war’ is over and the industry is settling on Kubernetes (K8s) as the standard.
However, picking the core technology (K8s) is only the easy part, there are lots of choices left:
- do you pick a fork or a vanilla Kubernetes?
- how do you handle upgrades, given there are ~4 Kubernetes releases every year?
- how do you handle multi-tenancy?
- what about security?
The easy way out is picking a hosted version on one of the hyperscale clouds, as it takes care of some of these questions. Every vendor has its own flavor, with one common trade: they all have different management APIs and GUIs that are specially engineered to make it hard to move away from them. And so, while the containerized workloads themselves are portable, this will make it hard to drive a multi-cloud strategy.
So what now?
VMware PKS answers all the questions above, without the cloud vendor lock-in dilemma above:
- it’s a vanilla Kubernetes, which means you don’t have to wait for a vendor to upstream the latest changes
- it’s based on BOSH – a battle-tested open source tool for deploying and managing distributed systems – which handles the upgrades and availability questions
- it handles multi-tenancy with a multi-cluster approach: every ‘trusted domain’ gets their own managed K8s cluster
- networking is based on VMware NSX, in which containers are first class citizens, and which enables policy-based security
Since version 1.3, PKS can be deployed on every hyperscale cloud (Microsoft Azure, Google Cloud Platform, and Amazon AWS) as well as in on-premises VMware vSphere data centers, making it a true multi-cloud solution with consistent management, operations, security, and developer experience.