The Cloud is Much Cheaper!

The short answer is…it can do.

But as with Cloudbusting | The Cloud is more Expensive!, only if you put in the effort and forethought to make it as cost-effective as possible

Just deploying resources in the Cloud will almost certainly be more expensive in the long run

It will be, if you…

Use Budgets

• Budgets are non-enforced ‘caps’ on spending; they offer the ability to set spend thresholds based on either static or dynamic criteria that will alert key stakeholders if the budget is close to or being breached
• You can set multiple thresholds, depending on what you need to get out of the alerting

Use Policies

• There are a lot of expensive resources in Azure; most that you will never need (the GPU VM is cool, but unlikely we’ll be doing any video editing)
• You can use policies to prevent users from creating resources based on any criteria you can think of; referring to size here but you can limit redundancy options, region, you can set the required tags (tagging can be very important for reporting and budgeting) etc.

Minimise cross-region traffic

• Inter-region network egress is chargeable in Azure; if you set up your VMs in West Europe, then build a logging tool in North Europe and send all of your diagnostic data there you’re going to have a bad time.
• It’s really important when designing your infrastructure to minimise the amount of cross-region traffic
• One area that we don’t follow this is SQL redundancy; we often recommend multi-region failover clusters to improve recoverability in the event of a regional failure

Reserve Instances

• Reserving your resources can offer savings of up to 65% compared to the pay-as-you-go price
• You are committing to Azure that you will be using this size or family for the given period, which allows them to better forecast and in return you get a substantial discount
• That being said, if your needs change a lot of reservations can be refunded or exchanged for new ones if needs be (perhaps that new size family comes out that you desperately want)
• It’s also available for more than just VMs, a number of Azure resources can benefit from reserved instances

Forget about the basics

• By default, a lot of redundancy is disabled by default.
• There is a lot available; Local, Zonal, Global across different resource types; but the more redundancy you build in the higher the cost
• Similarly, in a lot of cases backups are not on by default and will incur extra charges
• And advanced security features in Azure AD, such as Conditional Access policies, require at least an Azure AD Premium 1 license.
• There is no hard and fast rule on what you should and shouldn’t enable; only you can determine the best fit for you and your infrastructure.
  • For example, in our multi-region SQL cluster we might create the VM with Local Redundancy (3 copies within the same datacenter) because we have a secondary node in a completely different geographical region

Minimise on performance

• It’s really easy to make it cheap if you go for the lowest performing resources.
• There’s again no hard and fast rule here; it’s going to depend on your workloads
• We would say using HDD is often not worth it, Standard or Premium SSD at a minimum
• Equally, the B-Series is great for jumpboxes and low-transaction systems but would not recommend for most production use cases

If you want to find out more about our Azure Managed Service or Azure CSP, please reach out and we’ll be happy to start a conversation