Cloud is a business strategy, not an IT implementation detail
British Airways is not having their best weekend: A big IT outage, resulting in a lockdown at Heathrow, all apparently caused by a single power switch. While I don’t know the details of their systems, they appear to be running everything on-premise. I am convinced that a true cloud-native approach, with multiple data centers, high availability, auto-scaling, chaos-monkey testing, infrastructure as code, PaaS, FaaS, … can avoid such a huge PR, business, operational and financial failure. BA, and many other businesses, cannot operate without their IT systems being live. It’s not an efficiency drop. It’s not an inconvenience. Your operations stop. Clearly IT infrastructure is core to how most businesses operate. And it is not acceptable to stick with yesterdays technologies in todays world. Yesterdays technologies were never designed or built for this kind of scale, or this kind of availability.
Yes, cloud has its own issues. No, it’s not because you’re in the cloud that all your downtime disappears. But the cloud does give you the tools and capabilities that no on-premise datacenter can offer you. And the term “private cloud” is mostly meaningless. It’s a rebranding of the on-premise data center. Here are the questions you need to ask to know if your “private cloud” is really a cloud:
- Can I truly scale my compute power on-demand? The answer is probably no, because you are limited by the number of servers in your data center, and by the bandwidth of your expensive SAN.
- Am I truly resilient against (inevitable) failures? The answer is probably no, as was clearly the case with BA. The shocking thing is not that the power supply went down. The shocking thing is that it takes hours (days?) to recover from a failure.
- Can I truly leverage technologies “as-a-service” without getting bogged down in maintenance, patches, backups and security updates? The answer again, is probably no. S3, RDS, DynamoDB, Lambda, EMR, … and their counterparts at the other cloud providers, are all super powerful technologies, yet require almost no effort from the end-user. Nothing on-premise comes close to this.
In the cloud you have more storage and servers than your credit card can handle. You can run hundreds of tests spinning up and killing machines, network interfaces or even entire data centers using tools such as Chaos Monkey. With PaaS and FaaS becoming more popular, you don’t even need to manage machines anymore. You can build a scalable, high-available, resilient cloud architecture almost “out-of-the-box”.
Don’t have the needed skills to do it yet? There is a thriving eco-system of service providers, consultants and freelancers to assist you with that. The good thing about the cloud is that the skills are transversal across companies. Also, many pioneering companies, such as Netflix, actively share their experiences with the cloud, both good and bad, at conferences and meetups. And as always, there are plenty of training options, certification programs and online courses to learn it all yourself.
So, next time you have a discussion about going to the cloud. Don’t focus on the cost of S3 vs your SAN, or the cost of EC2 vs your hardware vendor. Focus on the cost of a major outage and how exactly your cloud native architecture is going to reduce the risk of that ever happening. Focus on the speed-up in development because of self-service infrastructure and PaaS / FaaS offerings. Focus on the value of being able to release 10x because everything is automated, including creation and configuration of infrastructure.
And, as a final straw, if no argument under the sun can convince your senior leadership team of the value of the cloud, tell them they will have to explain the failure to the entire nation, wearing a funny security shirt, as Alex Cruz did. :D
Originally published at https://www.linkedin.com on May 28, 2017.