5 Cost Optimization Tips For Cloud Architects

These five cost reduction practices all have a common foundation: getting a tool to do the heavy lifting and then showing the Cloud Architect the cost data, in context, of your cloud architecture.

If you don’t have a tool that makes this easy then the Cloud Architect will be much slower doing this, more frustrated and find fewer opportunities for cost optimization.

It’s like a car engineer having a specialist diagnostic tool compared to one that doesn’t: who do you want to fix your car?

One will be fast and accurate and one will not.

The cloud-native tools might be free but they are “raw”.

Seeing a chart on a dashboard with the label “EC2” is useless, and drilling down to an unintelligible instance name isn’t much more use: where is this zombie EC2 instance, in which application, which zone? Who owns it?

You need to see cost information IN your cloud architecture.

Having costs in one tool, the cloud architecture diagram in another tool (or worse, statically produced instead of a live view), and the cloud console detail in another screen is a frustrating and error-prone experience for Cloud Architects.

The five practices explored are:

  1. Reduce the size of your cloud footprint by eliminating zombies
  2. Rightsize resources that are burning cash
  3. Set up automated monitoring
  4. Take a global view to see trends and anomalies
  5. Put alerts, warnings and budgets in place

You can achieve all of this quickly, simple and easily out of the box with a 14-day trial of Hyperglance.

1. Reduce Your Cloud's Footprint by Eliminating Zombies

This practice answers the question: How do I find Unused or Unattached Resources?

Remember you pay for what you order, not what you use -- unless you are using serverless compute like Lambda or serverless databases like Aurora serverless where you pay per transaction.

Did you know that even when you stop an EC2 instance you’re still paying for that instance’s EBS storage? How do you know which of the hundreds or thousands of EBS storage meet these criteria?

So, in a nutshell, you need to buy things that are the correct size and turn stuff off when you’re not using them. This is a tedious process for humans so you need a decent tool to do the scanning for you and send you notifications of cost reduction opportunities.

Examples of “unused or unattached waste” are things that are spun up for a test then forgotten and left live but unused later. Think unattached IP addresses, for example, that cost you money that adds up quickly.

Hyperglance not only looks for these but it links discovered zombies directly to your cloud map much like Google Map overlays restaurants and petrol stations and lets you interact with them.

A zombie load balancer discovered by Hyperglance
Hyperglance showing the zombie load balancer on a map - the instance is not active

2. Rightsize idle resources that are burning cash

This practice answers the questions: 'How do I identify and Consolidate Idle Resources?' and 'How do I right-size computing services like EC2 instances?'

It was common in the old on-premises world to have virtual machines running at 3% CPU capacity; they “felt free” and the cost wasn’t obvious.

In the cloud, costs are obvious and painful.

If you repeat the on-premises behaviours of just deploying virtual machines that do one thing only (e.g. webserver) and they are lightly utilised, you need to either reduce the size of the VM or combine multiple services.

The cloud architect is key here to make sure the design accommodates this.

An old-world, on-premises architecture will cause this cost wastage.

By looking at your cloud map in Hyperglance, these wasteful resources are highlighted in context by overlaying their performance on top of the instance in the cloud architecture, making the cloud architect's job much simpler.

Hyperglance automated advanced search rules for idle EC2 instances

3. Set up Automated Monitoring

A Cloud Architect should spend their waking hours watching a dashboard or manually searching for savings.

It’s essential that they can program an automated “extra member of the team” to always look for this waste.

That is, every time a new kind of waste is found, a rule should be created to always look for that waste from now on.

This gives the Cloud Architect the confidence to say in a meeting, “We not only discovered this waste and eliminated it once, but we will now avoid it in future because we’ve taught the system to look for it.”

All of the Hyperglance dashboard items are automated rules engine checks.

Hyperglance Automated Search for Idle Gateways

4. Take a Global View To Find Anomalies

If you only run services in the US and you spot a large high-performance EC2 instance mining bitcoin in Asia then you might have a problem.

What about if you run services across more than one cloud?

The Hyperglance Cost Explorer makes it easy to see what’s happening in places other than your commonly used zones.

Without this “global map” and using cloud-native tools you’ll have to purposefully flip from region to region to check for usage - this can be a painfully slow process and, for that reason, people tend to put it off.

First, you can see two clouds are in use across many regions and zones in the Explorer interactive map:

Hyperglance AWS cost explorer

By clicking on these items you can explore the map interactively.

For example, by clicking on the Amazon EC2 cloud resource on the right, I can see which regions and accounts where that resource is being used:

Hyperglance AWS cost explorer

If you only ever use the US regions and non-US regions appear… this is a great way to spot rogue cloud use and trigger a Cloud Architect to disable cloud regions in future.

5. Codify Your Cloud Cost Policies

Cloud cost guardrails are essential.

The best way to create them is to code your cloud policies into automated engine rules.

For example, you might have a rule to eliminate orphan EBS snapshots that are over 30 days old.

This can be codified like this:

Hyperglance automating discovery of orphaned EBS volumes

Finding this will probably uncover an inefficient cloud practice of people not cleaning up after themselves.

This kind of rule lets the Cloud Architect implement the “Trust but Verify” practice: trust people to clean up, but have a safety net for when they don’t.

Hyperglance & Cloud Cost Optimization

If you're looking to improve your cloud cost optimization, Hyperglance is the perfect place to start:

 

  • Hyperglance is shipped with hundreds of built-in & customizable checks, all designed to help you enforce policy and reduce your cloud bill
  • The checks run continuously, and are based on best practises and industry frameworks (CIS, NIST, NIST 800-53, NIST 800-171, AWS Well-Architected, HIPAA, PCI DSS, & FedRAMP)
  • Checks can be used to trigger notifications, and automate fixes using AWS SNS or Azure Event Grid
  • Hyperglance is self-hosted, deployed  through the AWS & Azure Marketplaces, in Kubernetes, or installed on your own instance/VM