We try to help our customers understand how to manage costs within Azure, but there can always be surprises if we aren’t careful. Here are a few of the items that can impact your cost in a negative way:
- Leaving a deployment on when it isn’t needed. While this may sound funny, the reality is that most deployments are only used during the business week, and usually have schedules to take them offline after hours. The concern is actually the weekend, what if someone wants to work on a Saturday, but neglects to turn off the deployment when they are done? The deployment could remain running the rest of the weekend. My solution is to create an additional schedule item that turns everything that needs to be turned off every evening (including weekends) at a time when you know everything should be turned off.
- Playing “What if” with a deployment and not cleaning up your scenarios after you test them. I spend a lot of time trying different scenarios, and find that if I’m not careful, I’ll leave VMs running long after I’m finished with them. This is another place where I use the schedule. All my “test scenarios” immediately receive a Shutdown scheduled event every evening at 7:00 pm. This way if I forget to turn them off, they will be taken care of. The other side benefit of this is how many times do you realize you forget to do something as you are winding down for bed? Now, I can sleep easy knowing my schedule has my back.
- Users that consume more storage than anticipated. While Standard Storage in Azure is relatively inexpensive, $50.00 / TB / month, the same amount of Premium Storage can cost almost three times as much. Adding storage does require that someone manually allocate the additional storage but forgetting to account for it can be a frustrating mistake.
- Leaving unneeded VPN gateways running. If they aren’t needed, and aren’t going to be used, remove them because they have an hourly cost as well. You may be running a VPN gateway if you started with an on-premises infrastructure that you’ve since migrated to the cloud. If the VPN gateway is no longer needed, delete it and save the additional cost. Note, that if you must re-create a VPN gateway, it will be created with a new IP address. So, if you feel you are going to need the same gateway in the near future, it is probably less hassle to leave it running than reconfigure the end points depending on the VPN infrastructure.
- Not leveraging scheduling or auto-scaling. If your customers aren’t using all or part of their deployment during certain hours of the day or week, turn off all or part of your deployment. We have an easy User Interface that will walk you through setting up the scheduler for one, a few, or all VMs in your deployment. Click here to learn how to set up the Scheduler feature.
I really like our auto-scaling capability, it gives your infrastructure to ability to scale based on user demand, then after hours scale down your infrastructure as users sign out at the end of the day. This is a great configuration to ensure you have the available capacity when needed, but that you are not paying for maximum capacity when your demand is minimal.
- Assigning users that do not need RDS access to the RDS collection(s). You are billed for the MyCloudIT license and the RDS SALs, if purchasing RDS SALs, based on the number of users that have access to your RDS collections whether the users access the collections or not. This is how RDS licensing works, and we have adapted our licensing model to match the Microsoft model. If you add unneeded users, or user groups like All Users to the RDS collection, means you will be charged for all users that have access to the RDS collection(s).
As I talk to customers and partners, one of their biggest concerns with Azure is the variable cost. We provide tools that allow you to take advantage of that variable cost model to reduce your overall runtime costs.