Jan 17, 2020 Why Cloud First Design? 10 Points in 10 Minutes
Yesterday I moderated a panel discussion at a Wall Street Technology Association event on cloud-first design. I was asked to give an industry perspective to lay the ground work for our wonderful panel of technologists. With 10 minutes and no slides, I offered these 10 points to consider in support of pursuing a cloud-first design philosophy.
- Cloud is everywhere. Overall, more than two-thirds of organizations now make use of IaaS and PaaS, and among financial services companies it is more than 79% — with a bullet. By end of year, our research says more than 95% of financial firms will be using IaaS and/or PaaS in production.
- Cloud is now dominant. In 2019, for the first time, more than half of the work of the average enterprise was being done in IaaS/PaaS or SaaS; only 40% remained in company data centers. For financial firms that number is even lower, about 37%
- Most folks don’t save money going to cloud. Using IaaS and PaaS is still coming at a roughly 12% premium overall, on average, compared to deploying in-house. However, most organizations no longer cite saving money as the main reason for shifting things to cloud.
- They do cite agility, scalability, and resilience. On this score, cloud is delivering well, lowering the barriers to trying something new, and significantly simplifying scale-up (and scale-down) on successful solutions.
- To get all those other benefits (and to get them to an even greater degree) and save money, you need to deploy solutions that are designed to run in a cloud, to take advantage of cloud’s virtues (like dynamic scaling), and to behave like cloud-native applications by autoscaling, being geographically resilient, etc.
- Microservices is a major feature of most cloud-native apps: factoring an application into relatively large numbers of relatively small functional units. About 40% of organizations have at least begun to design with a microservices approach, although for most it will be a multi-year effort to make it the main model. Containers go hand in glove with microservices (although they are also useful to non-microservice applications, of course); already more than 55% of organizations use them, and more than 75% will soon.
- Using DevOps techniques is another major factor: with an expectation of rapid initial delivery of an app becoming the norm, and rapid evolution thereafter (see the next item), steadily more organizations embrace agile and DevOps approaches such as scrums and sprints and heavy automation of all aspects of functional and security testing. Adoption of most of these was around 40% in 2019, and by the end of 2020 will be closer to 75%.
- With rapid delivery and continuous evolution the goal, CI/CD–continuous integration/continuous delivery–is rapidly becoming the default mode of operation. Already more than 45% have begun to adopt it (that is, for at least one project or program) and more than 85% expect to by year’s end.
- Really harnessing cloud as a platform and cloud-native/cloud-first as a design approach requires organizations to change not just design and technology but also team structures, position descriptions, and not just work processes but also those for hiring and for evaluating performance and for setting pay.
- Going cloud-first sustainably, especially with the shift to microservices and containers, requires a shift in mindset as well. Most IT operations and security teams came of age with dedicated physical machines or virtual machines as the dominant unit of work in a data center: entities that for the most part stayed put once created and that lasted for a long time, months to years. Most had human hands involved in their provisioning. With the shift to dynamic, horizontal scaling based on containers rather than virtual servers, the environment will be filled with things called into being by automation and typically existing for a very short time: seconds to hours. Monitoring and security systems have to be able to keep up, of course (hello AIops!) and staff need to be thinking with ephemerality and mutability as their new frame of reference.