Cloud Different

A new architecture for a new platform

Posted by David Dooling on October 6, 2015

Over the past several years, cloud computing of all types has gained a lot of momentum. Because of the massive increase in mobile computing, i.e., phones and tablets, more and more people are consuming cloud services like GMail, DropBox, Instagram, and Office365. These Software as a Service (SaaS) solutions are convenient for people on the go because the information they need is always easily accessible through a web browser or mobile app. This is possible because the user’s data and preferences are stored “in the cloud”, using servers and services managed by someone else, typically the operator of the SaaS solution. While these SaaS solutions are convenient for end users, they do not allow for much customization. To develop custom capabilities in the cloud, you must turn to Infrastructure as a Service (IaaS).

IaaS consists of virtualized compute, storage, and networking managed by someone else that you can consume, typically on a pay-per-use basis. The automation enabled by consuming infrastructure via services allows you to spin up an entire data-center-like environment in less than ten minutes! Many IaaS providers like Amazon Web Services, Microsoft Azure, and Google Cloud Platform provide additional capabilities like managed message queues, notifications, and even databases delivered as a service. Using these building blocks of IT, you can rapidly create solutions for your customers without purchasing a single piece of hardware.

So what’s the catch? While self-service infrastructure and paying only for what you use sounds great, typically cloud infrastructure is less reliable and over time becomes more expensive than on-site resources. To reap the benefits of IaaS, you must think differently about your solution architecture. If you don’t, you’ll end up with a less reliable, more expensive solution.

Cloud Different

To avoid these pitfalls, when architecting solutions in the cloud, you must design with the following five principles in mind.

  1. Automation: You must automate every aspect of your infrastructure and solution, ensuring your solution can be spun up reliably, rapidly, and repeatably.
  2. Fault Tolerant: Since cloud infrastructure is less reliable, you must create solutions that are fault tolerant, responding to and recovering from failures automatically.
  3. Horizontally Scalable: Your solution must be horizontally scalable so it can scale up to meet demand and scale back down to control costs.
  4. Secure: You must secure every aspect of your solution environment. Security is your job, not someone else’s!
  5. Cost Effective: You have to use commodity, cost-effective components so that executing on the first four principles does not break the bank.

Done properly, cloud solutions can be more reliable and less costly.

Just as how you architect solutions in the cloud must change, continuing to use the technology and processes that have been optimized for internal operations is not going to lead to success in the cloud. At Monsanto, when we asked ourselves how we could be successful in the cloud, we knew our people would have to adopt new technologies and develop more agile processes. We would have to transition from different teams managing different part of a custom applications life cycle, i.e., one team for development, another for testing, another for operation, and yet another for support, we should have teams operating in a DevOps model, owning a capability from cradle to grave and providing it as a service to their customers.

The cloud presents an amazing opportunity to increase the speed with which companies can deliver capabilities to their customers. To capitalize on the opportunity, you must embrace the cloud’s differences and use them to your advantage. The architectural principles outlined above coupled with more agile process and a holistic DevOps model allow teams to create solutions that are more secure, more reliable, more performant, and more cost effective than on-premises solutions. In addition, your teams will be able to deliver solutions that delight your users much more quickly than before.

To learn more about how to architect “cloud first” solutions, learn about our journey to the cloud, and see how we use some of the tools and libraries we have open sourced, be sure to attend our talk at AWS re:Invent, Cloud First: New Architecture for New Infrastructure, this Thursday 8 October 2015 at 2:45 p.m. PT in Palazzo N.

posted on October 6, 2015 by
David Dooling