A platform is more than infrastructure and APIs – it is how it is implemented in a way that empowers delivery teams to use it and the way that it supports them to do so. It is a complete set of capabilities that need to be architected, tested and deployed to allow application teams to self-provision their infrastructure with a range of sensible options, including integrations, templates, support and evangelism. It is a challenge to get this right, especially with a large organisation steeped in traditional infrastructure and operations thinking, but it is a critical element of a successful DevOps implementation.
A common analogy is the platform workshop that builds a chair, but uses that same foundation to construct tables, sculptures and hundreds of other things. This is what a good platform does – it provides the underlying capabilities that a number of different offerings can build upon – but it also ensures that those elements are secure, scalable and extensible.
A platform business doesn’t directly create and own its inventory in the same way that linear businesses do – it simply acts as the means of connection for the supply chain. This allows it to reduce transaction costs and facilitate innovation from the outside rather than within its walls. This is why platforms are such an important element of our networked economy and a reason that we should all understand how to design them. For example, Datadog is a platform that gives its users the ability to monitor all of their infrastructure and applications, including performance and logs. This can enable them to track issues and problems across multiple services and teams, from one central location, so they can identify and address them in a timely manner. platform