By Meg Ramsey
CEOs and boards are demanding that IT go all in on cloud. Gartner believes that if you haven't developed a cloud-first strategy yet, you're falling behind your competitors.
But not all clouds are created equal. You need to choose the right infrastructure based on your application workload's needs, and should pursue a multi-cloud strategy that not only takes advantage of providers' different strengths, but also prevents vendor lock-in.
How do you create a strategic buying framework for a multi-cloud strategy that takes into account all your business, system, application and cost drivers?
I'm going to walk through the full process in a series of posts covering:
- How to develop a cloud-first strategy, gain executive buy-in and create governance and compliance requirements
- Identify your existing application landscape, create criteria for application destinations, and perform application workload placement analysis
- Begin migrations and application refactoring, and create a feedback loop to periodically evaluate for efficiency
But first, I want to look at the current data on types of cloud infrastructure and individual providers. Each is suited for different applications, and by starting with an overview of the market right now, you can more strategically and successfully approach the cloud.
How to pick the best cloud solution for each application use case
I created a chart based on research from Gartner, IDC, and Forrester that gives a high-level summary of what each type of infrastructure works best for and what you should probably avoid.
- On-premise data centres are great for latency-sensitive workloads or to address privacy or data security concerns, but compared to other services from public cloud and SaaS providers, that's where the differentiation ends.
- Colocation is using another data centre provider's space and power. This is good for legacy applications not ready for refactoring to run in the public cloud.
- Managed hosting is the same as colocation. It's good for legacy applications that aren't ready for refactoring. However, if any of these services are undifferentiated and could be replaced with SaaS tools, you should consider moving them to SaaS instead.
- Hosted private cloud (HPC) offers great value for enterprise mission-critical workloads that are steady and don't need elastic or burst capabilities. You can still scale on demand, just not to the same volume as public cloud providers. HPC is a good option if you have applications that are finicky around noisy neighbors and require a single-tenant environment, or if you have data, security or privacy concerns. If you have elastic workloads or ones that need to scale rapidly, you should look to the public cloud.
- Public cloud offers highly innovative services with scalable and elastic infrastructure that allows you to scale both horizontally and vertically depending on your workload's needs. However, it can be costly to lift and shift an application from a virtual environment to the public cloud, so be sure to optimise that workload for public cloud services. Steady state workloads have the potential to be more costly in the public cloud. If you move steady state workloads to the public cloud, make sure you take advantage of all the cost-saving services, like reserved instances or the spot market, and that your application takes advantage of higher level PaaS services only offered in the public cloud.
- Software as a Service (SaaS) is best suited to replace off-the-shelf applications you've been running and maintaining for years. This is a great solution for where you're not differentiated and just need tools to run that part of the business, freeing the IT team to focus resources on revenue-generating opportunities. That's why SaaS isn't best for differentiated services, especially areas where you've invested a lot in intellectual property, unless you're creating your own SaaS tool to sell to customers. Those applications are best suited for public and private cloud environments.
Applications should drive infrastructure decisions. It's important to ask if an application truly needs to leverage the characteristics of the cloud and create a framework that determines where applications should sit from an infrastructure standpoint.
Keep in mind that as you move from on-premise data centers to SaaS, you increase vendor lock-in. Most analysts recommend that you have at least two cloud providers, with 70-80 percent of workloads residing in your primary cloud, but building relationships and familiarising your team with the second cloud to create a secondary buying path if necessary.
How to choose the best cloud for your applications
Not all clouds are created equal. I pulled together the chart below to offer a high-level overview of the different cloud providers out there, the ideal use cases for each and what you shouldn't use them for.
We'll dive deeper into this chart in future posts.
Getting started with a multi-cloud strategy
The infrastructure and providers you choose will always depend on your use case. In future posts, we'll look at how to apply this information to winning executive buy-in, honing in on the best solutions for your applications and beginning migrations and refactoring.
In the meantime, let me know if you have any questions or want to discuss your particular cloud strategy.