Cloud infrastructure has been around long enough that IT professionals like myself have several stories to tell and a few battle wounds to show. In this article, I share the lessons that fellow customers, colleagues, and I have learned as we cut our teeth on migrating to both public and private clouds.
If you have experiences you would like to share, we invite you to post them in the comments below.
Good Idea: Using Pre-Purchased / Reserved Instances
Initially, most organizations leveraged the subscription or pay-as-you-go model. This approach allows you to be charged based on your actual utilization and varies in cost, much like a regular utility bill. Ultimately, this provides a lot of agility and helps lower the barrier of initial entry. As utilization increases, however, the cost associated with leveraging the pay-as-you-go model long term is far from ideal.
To compensate, cloud providers have put contract vehicles in place for providing resources at a drastically reduced rate. By pre-purchasing resources in yearly increments business can save upwards of 65% on their cloud computing overhead. Most organizations opt for this choice and use the pay-as-you-go model when a situation calls for a short-term environment.
Bad Idea: Plan to migrate everything at once
Migrating to a cloud, public or private, can be a difficult process that requires adequate time and planning. It’s understandable that IT teams are in a hurry to move all their workloads as soon as possible so they can leverage the features that the cloud offers. However, it’s paramount that adoption is taken slowly and done with care.
It can be complicated to estimate the financial overhead of running workloads as the cost heavily depends on storage, networking I/O, instance sizes and more. Plus new tools and practices will need to be incorporated into your organization. You’ll want to have some experience and a few small wins under your belt before you take on migrating your larger and more critical workloads.
Good Idea: Architecting your systems to leverage the cloud
It’s not uncommon for organizations to mistakenly think they can “lift & shift” monolithic applications into a cloud and that it will magically become scalable. The reality is the underlying infrastructure might scale, but your monolithic application won’t unless it’s built to leverage failover, load balancing, session persistence and more.
Plus, the bulky nature of monolithic apps will cause your cloud consumption costs to be far higher than is ideal. And some applications, like those with heavy I/O may be better off staying in a private cloud or co-location to stay performant.
If your application wasn’t designed for a cloud environment, consider updating it so it can perform optimally in a cloud environment, or use an alternative infrastructure like an on-premise server, co-location, or private cloud. This may go against the modern day hype, but you don’t have to move every application to the cloud.
Bad Idea: Assume all products (and licenses) are cloud supported
You’ve done all the work and made sure your application is ready to be deployed to a cloud environment but sadly… you forgot to check if the software your application leverages is cloud ready.
Or to add insult to injury, maybe the products aren’t supported for cloud usage. This happens more often than you might think. Before architecting your system, make sure all the underlying components are cloud ready as well to avoid a headache later!
Good Idea: Settling contractual obligations for meeting SLAs
Before signing on the dotted line when adopting a cloud provider, be sure to carefully review the details. You will want to pay close attention to the SLAs and operational level agreements. Not only should they be well defined but they should be strict enough for your business.
This is one of those details that frequently gets missed and won’t come up until an incident occurs. When systems go down, you want a strongly written agreement of how your organization will be compensated for lost service as it impacts your customers and ultimately your business.
Bad Idea: Making security an afterthought
Coming from a security background, I’m always shocked at how many organizations leave security and risk analysis to the end. The fact is moving to a public cloud means you and your customers’ data will likely be hosted online. This can be especially concerning if you work in business vertical that has tight regulations (i.e. HIPPA, Sarbanes-Oxley, PCI, etc).
A lot has changed over the years and multiple cloud providers are touting the security certifications they currently hold. This is great news, but the work doesn’t end there. During your architecture and design phase, set aside time to consider any security holes that might exist as these problems may persist for a long time. If there are changes to be made, now’s the time to make them.
Good Idea: Consider a multi-cloud approach
It’s rarely good wisdom to put all your eggs in one basket. Especially when that basket is the backbone to keeping your organization up and running. Even if you have a great data recovery strategy and will be properly compensated for infrastructure outages, time lost cannot be regained. Any system outages can impact your income and potentially even worse, your brand and corporate image.
A few times last year, we witnessed public cloud outages that ranged from contained instances all the way to covering an entire coastline. It’s when these situations arise that you ultimately realize that you should always be prepared and have redundancy across providers.
Implementing your workloads across multiple cloud providers is called a multi-cloud approach. Not only will this help protect your corporation, but ultimately it will instill a modular platform design and prevent potential vendor lock-in.
Moving to the cloud is exciting and is a top priority on the IT roadmap of many organizations. There are literally dozens of things to consider when selecting a cloud provider and migrating your applications. We hope these tips gives you a few things to think about as you prepare to make the transition and help you avoid a few of our battle scars. But know you don’t have to go it alone.
From blog articles, webinars, events and outsourced support, you have many resources to successfully make the transition for your organization. If you still have questions after spending some time researching, shoot us an email. We are happy to help.