More than 85% of enterprise organizations today are leveraging the cloud to some degree. Regardless if you’re a new Azure cloud adopter or a mature, native user, finding success in the cloud isn’t easy. So what challenges are organizations facing in the cloud today? If you’re a large enterprise, lack of collaboration between lines of business is probably a big one. Are you actively working to ensure all stakeholders have bought into a single, cohesive azure cloud management strategy? If your answer is no, this is a great first place for your team to start.
Already have everyone onboard? Great—the work doesn’t stop there. Once you’ve established an Azure cloud management strategy, you need to work on tying those strategic initiatives to concrete business objectives, a difficult task for large organizations where departments are making decisions independently of one another.
Explore all the Azure best practices from our team of experts in our full webinar recording.
We put a bunch of Azure experts in a room and asked them to share tips and tricks on how you could better scale your cloud infrastructure and keep your environment continuously optimized. What did they come up with? A ton of best practices on how to eliminate waste in your Azure cloud environment, how to reduce and optimize costs with reservations, and ways to centralize governance and save time with automated policies.
Here’s what they had to say about developing an Azure cloud management strategy.
It’s very common when you’re deploying workloads in Azure to oversize or over provision specific workloads, says Michael Mouchantat, Solutions Engineer at CloudHealth Technologies. Before you start rightsizing, make sure you understand how your machines are performing. For your virtual machines (VM) you should be looking at CPU, memory, the disk, and the network. For SQL, looking at DTUs (Database Transaction Units) and storage will provide important insight. If you have a compute optimized VM that’s only using 10% or 20% of the available CPU capacity for example, you have an opportunity to change that machine type to something that has less CPU capacity or potentially terminate the workload.
Below are some common causes of unused infrastructure and how you can fix it:
Reservations can supply you with up to a 72% discount on your virtual machine and SQL Database costs compared to pay-as-you-go. If you have virtual machines or SQL Databases that runs for a long period of time, purchasing a reservation is one of the most cost-effective ways you can take to reduce your cloud spend. Just be aware that when reserving virtual machines and SQL Databases, the reservation discount only applies to the compute costs, not any other associated costs — software, networking, and storage are all billed separately.
Check out the chart below comparing Azure reserved virtual machine instances and SQL Database reserved capacity:
When a customer asks how much of their infrastructure should be covered by reservations, Samir Mehra, Director of Products at CloudHealth Technologies, says a best practice is not let it dip below 60%. For reference, some of CloudHealth’s most mature customers have about 80% of their Azure infrastructure covered by reservations.
Still not ready to commit to a long-term reservation? You might have the option to make your reservation size flexible, meaning you have the option to modify your reservation down the road to ensure your Azure environment is optimized for your infrastructure’s needs.
Why should you be thinking about establishing proper cloud governance? Think of it as building guardrails within your environment that ensure secure, effective, and efficient use of your resources. It also acts as a control system to prevent others from working outside their scope of authorization inside your environment.
Marie Burke, Sr. Product Marketing Manager at CloudHealth Technologies, says it's all about getting adjusted and comfortable with defining policies to take action within your Azure environment. To start, ask yourself how many of your virtual machines and SQL Databases do you really need to run 24/7? For example, some of your machines running in QA might not need to run on the weekend. An easy solution is to define and leverage policies that spin down non-productive assets on Friday night and spin them back up Monday morning. You can automate this policy to ensure you aren’t paying for unnecessary compute.
Check out what Burke has to say about the importance of establishing policies.