With all the talk of containers, cloud-native applications, and cluster management platforms, it’s assumed that the enterprise migration to the cloud is a foregone conclusion. I’ve wanted to provide a little perspective. Some perspective has come from the poster child for all cloud infrastructure, Netflix.
Netflix recently published a blog post championing their seven-year journey to the public cloud. Netflix has contributed significantly to the overall knowledge of operating a predominantly cloud-based service but, it’s only after years of preparation that they were able to migrate the streaming portion of their service completely to the public cloud. Here are five top takeaways from their journey.
1. It’s not about cost savings
One of the impressive attributes of the Netflix cloud use case is the clarity around the value of cloud. Netflix didn’t communicate cost reduction as an advantage. Instead, Netflix spoke of the advantages of scale and reliability in leveraging AWS. Netflix realized cost savings due to the elastic nature of their workloads. The company wasn’t burdened by the fixed cost of scaling their private data center for peak load. However, by building elasticity into their application, they were able to reap the additional benefit of cost savings.
SEE: Cloud Data Storage Policy Template (Tech Pro Research)
2. Change management
Undertaking a cloud migration is very different than a traditional data center migration. Abstracting away the infrastructure also means that you give up some insight into the infrastructure itself. In the case of the outage that occurred Christmas Eve 2012, routine AWS maintenance on elastic load balancers (ELB) led to a disruption of the streaming service. While the problem isn’t unique to a service hosted in the public cloud, it highlights the challenges associated with migrating controls such as change management. An organization that has a mature configuration management system that evaluates the impact of each sub-system on a distributed system should take note.
It’s tough to achieve five nines (99.999%) availability in a private data center. It’s even harder to achieve that level of availability in a cloud-based application. Netflix has finally reached a four nines availability. In the past, I’ve used the rule of thumb that applications with two to three nines are targets for cloud migration. Just as Netflix was selective in the services they migrated to the public cloud, it’s critical that organizations understand the impact of service availability when migrating to the public cloud.
One of the reasons Netflix was able to save money on public cloud vs. the private data center is their application architecture. Netflix didn’t simply lift and shift monolithic applications from their private data center to an AWS VM. Taking the forklift approach ticks off the check box for cloud without any of the real benefits. Since scale and reliability were the primary factors in Netflix’s cloud decision, it required a re-architecture. Organizations considering cloud need to evaluate their current application architecture. Lift and shift to the cloud may just shift your existing problems to a new platform, sans the insight from managing your infrastructure.
Netflix is held up as the gold standard for leveraging cloud computing. The median annual salary at Netflix is an eye-popping $180K. There are not very many organizations that have the resources or reputation to attract the talent needed to undertake a full migration to the public cloud. It’s likely that most organizations will seek outside assistance in their migration efforts. The Netflix journey highlights the need to investigate deeply the credentials of organizations looking to assist in your cloud journey.
What are some of the insights you observed from Netflix’s cloud journey? Share you thoughts in the comments section.