We could dive right into a definition. But if you ask a handful of software engineers to define cloud-native, you will most likely get more than a handful of definitions. Is there no unified definition? Is it not mature enough for a concrete definition? Or maybe everyone has their own perspective, their own context. When we talk about a particular topic without consensus on the context then it is unlikely we will reach consensus on the topic. So first we need to define the context for our definition of cloud-native. It should come as no surprise that in a patterns book, we will start by defining the context.
What is the right context for our definition of cloud-native? Well, of course, the right context is your context. You live in the real world, with real-world problems that you are working to solve. If cloud-native is going to be of any use to you then it needs to help you solve your real-world problems. How shall we define your context? We will start by defining what your context is not.
Your context is not Netflix's context. Certainly, we all aim to operate at that scale and volume, but you need an architecture that will grow with you and not weigh you down now. Netflix did things the way they did because they had to. They were an early cloud adopter and they had to help invent that wheel. And they had the capital and the business case to do so. Unfortunately, I have seen some systems attempt to mimic their architecture, only to virtually collapse under the sheer weight of all the infrastructure components. I still cringe every time I think about their cloud invoices. You don't have to do all the heavy lifting yourself to be cloud-native.
Your context is not the context of the platform vendors. What's past is prologue. Many of us remember all too well the colossal catastrophe that is the Enterprise Service Bus (ESB). I published an article in the Java Developer's Journal back in 2005 (http://java.sys-con.com/node/84658) in which I was squarely in the ESB is an architecture, not a product camp and in the ESB is event-driven, not request-driven camp. But alas, virtually every vendor slapped an ESB label on its product, and today ESB is a four-letter word. Those of you who lived through this know the big ball of mud I am talking about. For the rest, suffice it to say that we want to learn from our mistakes. Cloud-native is an opportunity to right that ship and go to eleven.
Your context is your business and your customers and what is right for both. You have a lot of great experience, but you know there is a sea change with great potential for your company and you want to know more. In essence, your context is in the majority. You don't have unlimited capital, nor an army of engineers, yet you do have market pressure to deliver innovation yesterday, not tomorrow, much less next month or beyond. You need to cut through the hype. You need to do more with less and do it fast and safe and be ready to scale.
I will venture to say that my context is your context and your context is my context. This is because, in my day job, I work for you; I work alongside you; I do what you do. As a consultant, I work with my customers to help them adopt cloud-native architecture. Over the course of my cloud journey, I have been through all the stages of cloud maturity, I have wrestled with the unique characteristics of the cloud, and I have learned from all the classic mistakes, such as lift and shift. My journey and my context have led me to the cloud-native definition that I share with you here.
In order to round out our context, we need to answer two preliminary questions: Why are we talking about cloud-native in the first place? and Why is cloud-native important?
We are talking about cloud-native because running your applications in the cloud is different from running them in traditional data centers. It is a different dynamic, and we will cover this extensively in this chapter and throughout this book. But for now, know that a system that was architected and optimized to run in a traditional data center cannot take advantage of all the benefits of the cloud, such as elasticity. Thus we need a modern architecture to reap the benefits of the cloud. But just saying that cloud-native systems are architected and optimized to take advantage of the cloud is not enough, because it is not a definition that we can work with.
Fortunately, everyone seems to be in relative agreement on why cloud-native is important. The promise of cloud-native is speed, safety, and scalability. Cloud-native helps companies rapidly deliver innovation to market. Start-ups rely on cloud-native to propel their value proposition into markets and market leaders will need to keep pace. But this high rate of change has the potential to destabilize systems. It most certainly would destabilize legacy architectures. Cloud-native concepts, patterns, and best practices allow companies to continuously deliver with confidence with zero downtime. Cloud-native also enables companies to experiment with product alternatives and adapt to feedback. Cloud-native systems are architected for elastic scalability from the outset to meet the performance expectations of today's consumers. Cloud-native enables even the smallest companies to do large things.
We will use these three promises—speed, safety, and scale, to guide and evaluate our definition of cloud-native within your context.