Four ways to keep cloud costs under control

Control your cloud services – don’t let them control you

Cloud is the ideal solution for today’s unpredictable business environment. From supporting remote working to enabling organisations to get new applications to market quickly, it’s helped many survive and even thrive in recent months.

However, while the headline costs of cloud are appealing, there are hidden traps which have led to problems and overspend for many organisations. Once you have got rid of your in-house infrastructure, you have to rely on what the cloud provider offers. Portability between providers is not yet proven, and it is not in a provider’s interests to make it easy! Based on our experience, we’ve identified four areas to review to ensure you control your cloud services, rather than letting them control you.

Learn to decipher usage reports and billing

To avoid being caught out by unexpected costs, your team needs to become expert in deciphering cloud usage reports and invoices. As well as expected charges for servers and storage, there will be additional costs for ancillary requirements such as IP addresses, domain resilience and data transfers into, out of and between servers which need to be compared with usage and forecasts.

For example, an IaaS instance in AWS has five to eight metered costs that need to be tracked for a single Internet facing server. Azure and other public cloud services are comparable. Complexity increases if your organisation hosts complex, multiple server environments. If other elements are required to run the application, such as security, resilience, management, patching and back-up, these will appear as additional charges. This is less of an issue with SaaS, which usually has a standard per user per month charge, but with IaaS, and to some extent PaaS, other elements are ‘extras’.

Once you have worked out the origin of each cost, it becomes straightforward to identify the causes of any overspend. Now consider why costs have escalated.

Beware data transfer and egress costs

All public cloud services are metered, which has pros and cons. Controlling spending means understanding what constitutes usage and the impact of data egress costs. For example, in some applications servers have a constant two-way dialogue. If this counts as usage costs will quickly escalate.

In many services there are two costs: one per GB each time servers in different domains talk to each other, and a second per GB to send data over the Internet. For example, in AWS you are charged if you use a public IP address. Because you don’t buy dedicated bandwidth, there is an additional data transfer charge against each IP address – which can be a problem if you create websites and encourage people to download videos. Every time a video is played, you incur a charge, which will soon add up if several thousand people download your 100MB video.

The same applies with resilience and service recovery, where you will be charged for data traffic between domains to keep a second DR or failover environment in a different region or availability zone. Moving data between v-nets and regions and out of the public cloud also incurs a fee. Most companies that use a public cloud service pay this for day-to-day transactions, such as moving data from cloud-based storage to on-premises storage, and costs can quickly spiral as your tenancy grows.

Don’t pay for unused resources – but scale back rather than cancel

One of cloud’s greatest strengths is its flexibility. It provides what is effectively ‘pay as you go’ capacity – as usage changes, you only pay for capacity used. So the easiest way to start saving is to review resources and remove any that are not being used, as well as whether there are any unused licences. In our experience, most companies are over-licensed in many key applications.

If you have implemented an effective tagging system, you will have a clear view of the resources used for each business application. A key question is: “Are there instances that can be stopped because projects are paused?” For example, if you have furloughed staff, have you stopped the resources applicable to them? This could include projects they are working on, as well as virtual desktops and compute needs for applications. However, rather than cancelling services, our advice is to scale back those that are not being used instead. Cancelling a subscription means data is deleted and after 30 days this data cannot be restored.

Another tip is to audit all cloud usage and turn things off when possible to avoid paying for unused hours associated with overnight, weekend and annual working cycles (while being mindful of backup needs). Azure, for example, uses tags and built-in automation to allow resources such as virtual machines to be shut down overnight or when your business is not operational. You may be also able to reduce costs further if you know some of your instances need to be constantly available for at least a year, in which case they could be run as a fixed service for a fixed price using reserved instances at a substantial discount. However, please note that this only reduces the server cost, not storage and data egress charges!

Then consider whether you are paying to store old data. Deleting it reduces your spend and is good practice to ensure you comply with regulations and internal policies. Alternatively it could be migrated onto a cheaper, i.e. lower performance and/or less resilient storage tier. By following a proactive data management process, you can save money and reduce exposure in the event of a security breach. If you decide to archive data, there are cloud storage instances designed specifically for this purpose which will be more cost-effective.

Cloud is not like in-house capacity

One of the most compelling advantages of IaaS/PaaS is that it’s easy to spin up new services, containers or virtual machines. However, if you don’t keep a close eye on what is happening, costs can quickly spiral. There may be benefits to allowing departments to manage their own instances, but information needs to be integrated centrally to enable full cost analysis and ensure effective governance. Tagging instances and having well-planned tenancy administration and management policy and controls is vital to control this.

As one organisation we worked with discovered, buying new instances as if they were in-house virtual servers will quickly increase complexity and costs, and can be difficult to unpick.  It is therefore vital to understand how costs are calculated and choose options accordingly.

Finally, the pace of change in public cloud is very rapid; new and enhanced services are frequently introduced, and often not well publicised. To get best value you need to be prepared to regularly migrate your services between different instance types, classes or families. The cloud provider will not do this for you, so you will need the skills to do it yourself or contract a third party to do for you.