Making the most out of your cloud infrastructure

The cloud has lots of advantages besides cost-savings, but there can be no doubt that the possibilities for cost-savings are a major reason why many businesses are undertaking cloud migrations.

If that sounds like you, then here are some points you might want to consider.

Think seriously about going with a single cloud vendor

In theory, it can make a lot of sense to spread your business around different cloud vendors. This minimizes the possibility of disruption and eliminates the possibility of being tied to one vendor.

In practice, you may find that the process of undertaking even small-scale cloud migrations is a whole lot more hassle than it’s worth.

This is particularly likely to be the case for small disruptions. Additionally, one of the defining features of a good cloud vendor is that they have a straightforward process for assisting customers who want to move elsewhere.

If you absolutely must have 100% uptime, then you may need multiple vendors. You will, however, also need a cloud strategy to manage the potential cloud migrations.

Otherwise, the best way to reduce costs is usually to go with a single vendor. That way, you maximize your value to them as a customer.

This may result in perks such as volume discounts and enhanced customer-service options.

Get your sizing right.

The average cloud platform can have upwards of 1.5 million sizing options. These determine not only the size of servers but also the way they are optimized.

For example, servers can prioritize memory, storage capacity or throughput (and there are many other options).

Getting this right reduces costs and improves performance. It is, however, very difficult for humans to wade through all these options by themselves.

It is also massively time-consuming (and, of course, time is money). For both of these reasons, it makes sense to invest in Right-Sizing tools.

Clean up your unused and/or unattached resources

If you’re planning a cloud migration, then it’s a good idea to make sure that your cloud strategy has a process for making sure that basic housekeeping tasks actually happen.

If you’re already using cloud infrastructure, then check that this is actually happening and, if not, then update your cloud strategy to make sure that it does.

You may be surprised just how much you can reduce costs by simple housekeeping, like making sure servers are actually turned off when a job is done and/or that storage is removed from instances before they are terminated.

Stop the use of idle resources

Adding resources to a data center can be a real pain. Because of this, it often makes complete sense to leave a bit of spare capacity “just in case”.

In cloud environments this is both pointless and expensive. The whole point of cloud infrastructure is that it’s flexible and scalable. It’s also billed according to your CPU usage.

Even though an idle computing instance may only use 1-5% of the CPU power it needs when active, over time these little expenses can start to make a big difference to your bottom line and getting rid of them can therefore bring about significant cost savings.

Work to a schedule

Picking up from all of the points mentioned above, an effective cloud-cost management strategy will take advantage of what is possibly the single, biggest strength of a cloud environment and make sure that, insofar as humanly possible, resources are only used exactly when they are needed.

The way to do this is to create regular heat maps of peaks and troughs in demand. Over time, patterns will start to emerge and this can be used to create automated schedules to ensure that the right instance types are made available at the right times.

It’s recommended to keep generating heatmaps on a regular basis so you quickly pick up on any persistent changes in usage patterns and can take action to adjust accordingly.

Make the most of volume discounts

Amazon Web Services offers Reserved Instances and Microsoft Azure offers Reserved VM Instances, both are generally known as RIs and both essentially mean paying upfront to secure volume discounts.

These volume discounts can be as much as 75% and RIs are usually valid for at least one year, sometimes up to three. In other words, they are great news for anyone involved in cloud cost optimization.

Make the most of spot instances

Spot instances are basically the cloud environment’s answer to end-of-day deals in grocery stores, although they are typically sold on an auction basis rather than on a fixed-price basis.

Spot instances, as their name suggests, tend to be made available for short periods and typically need to be used within short periods, often on the same day.

This means that you want to avoid relying on them but you can really reduce costs by making astute use of them.

See Also

AWS Status