Cloud Cost Management: Where to start
This article was written by Rich Gibbons, ITAM Review, and Heather McKay, Flexera.
In the recent Flexera 2020 Digital Transformation Planning Report, 54% of organizations indicated Digital Transformation as a top priority.
As organizations digitize to transform their business and improve the customer experience, they are implementing disruptive technology to help achieve these goals. One of the key digital transformation technologies is cloud and it should come as no surprise that organizations are redirecting software investments from on-premises to cloud based applications.
While cloud investments help organizations innovate quickly, managing cloud costs has proven to be a challenge…and it’s something that will only continue to increase in importance.
Cloud cost management foundations
The Flexera State of Tech Spend 2020 report showed that IaaS/PaaS spend accounts for 18% of IT spend – more than double the spend on SaaS and over 80% of the on-premises software spend:
Their 2019 State of the Cloud report showed that cloud cost savings was the top initiative for survey respondents, and that focus on keeping cloud costs under control is still absolutely vital in 2020 and beyond.
The potential for wasted spend in the public cloud is huge, in many different ways and for many different reasons.
There is something of a divide in parts of the industry – on one side, those who say cost control is the most important element and on the other, those who place more stock in “traditional” ITAM practices such as managing license use, ensuring moving systems to the cloud doesn’t lose certain license benefits, tracking license compliance etc.
My personal view is that, while both are important, getting costs under control should be the primary focus for most organisations…simply because that is where significant savings can be made – and quickly. Equally, having an accurate picture of actual cloud spend will help the business correctly decide on the viability of other projects and initiatives.
Whilst it is hugely important, for example, to make sure an on-premises system can be migrated to the cloud without losing certain on-premises benefits – and thus increasing the cost – that shouldn’t be the first focus. By first of all implementing some relatively straight forward cost optimisation measures on your existing cloud environments, you can be saving a significant % of spend while the various teams discuss the migration for however many months that is likely to take!
Managing cloud spend is like building a strong foundation – get it right and all kinds of wonderful, long lasting things can be built on top…get it wrong and, in the long run, it will cost more and cause a whole heap of problems.
AWS v Azure
Although initially launched in 2002, Amazon Web Services properly launched in 2006 and Windows Azure (as it was then known) launched in 2010, giving Bezos et al. a 4-year head start – something Amazon took full advantage of. Flexera’s research shows that the gap is closing, with 67% of Enterprise respondents using AWS and 60% using Azure in 2019. This represents a Year on Year (YoY) 2-percentage point increase for Microsoft and a drop of 1-percentage points for Amazon.
While I’m not predicting the downfall of Amazon AWS, it is worth noting Microsoft’s long history of coming from behind to gain market dominance (as long as it’s not phones!) – just ask Word Perfect, Lotus and Novell. Microsoft Azure eventually becoming the overall #1 cloud platform wouldn’t surprise me at all.
Azure is also showing the fastest growth in customers with 100+ VMs in the cloud, with an 8-percentage point increase YoY compared to Amazon’s 2-percentage point increase. Overall, AWS is still leading with 33% of respondents using over 100 VMs against 25% doing the same in Azure.
How to stop the waste?
There are several steps organisations can take to reduce unnecessary cloud spend, and these are common across all the major public cloud environments.
Flexera’s State of the Cloud report shows the most commonly employed strategy is “rightsizing” instances (80%), closely followed by after-hours shutdown (71%), and requiring tags on resources (70%).
As the name suggests, this is making sure resources being created are the right size for the job. It’s just as easy to provision 128 cores and 512TB of RAM as it is 4 cores and 8GB of RAM…does that web server really need the same amount of processing power as the CERN?!
With cloud, it’s often said that you pay for what you use but more accurately, you pay for what you have turned on. If you turn off cloud resources that don’t need to be running – at night/at weekends for example – you can make significant savings of up to 60% quite easily. I explored it further in this Computer Weekly article.
Cloud platforms such as Amazon AWS and Microsoft Azure allow you to “tag” resources to group them together logically. Tags can be related to a certain project, team, office, cost centre etc. and give a quick way of making sense of all the resources being used. This can be used to measure actual costs against budgets and perhaps implement showback/chargeback policies.
These three are relatively straightforward as concepts, however, as is so often the case, the implementation can be quite difficult. One of the biggest problems is people and their propensity for forgetfulness. Forget to shut down the test environment on a Friday and that could easily be 60+ hours of wasted spend. Forget to tag those new SQL databases and that’s a chunk of cost that can’t be easily assigned to a project or department. What is one to do?
Cloud payment methods
Another aspect of cloud cost management is taking advantage of the discounts on offer. This isn’t as simple as it might sound – often just knowing what’s available and understanding how they work can be a hurdle.
Reserved Instances, where pre-paying for 1 or 3 years of a specific cloud service presents significant discounts, are available across Amazon AWS, Microsoft Azure, and Google Cloud Platform.
They require a detailed understanding of current usage – cloud and on-premises – in order to identify target workloads. Equally, the potentially mercurial nature of cloud hosted systems means you also need good insight into future plans to make sure a long-term commitment makes sense. ITAM must also be “in the loop” when it comes to system changes, both planned and otherwise, to avoid having Reserved Instances going unused, causing your cost saving efforts to instead contribute to wasted spend. AWS have also introduced “Savings Plans” – giving another option to consider.
Another thing to bear in mind is that the billing files from cloud vendors can easily run to millions of lines and can be almost impossible to parse manually.
To really understand what you’re being billed for may require a level of tooling, automation, and data science that hasn’t been required previously on-premises.
There are a variety of other methods available to help manage cloud spend but those discussed here are a great place to start. Particularly now, anything you can do to quickly reduce costs for your organisation is key – the ability to keep working, and keep innovating, is crucial – and ITAM can be a key driver of helping that happen.
About Rich Gibbons
A Northerner renowned for his shirts, Rich is a big Hip-Hop head, and loves travel, football in general (specifically MUFC), baseball, Marvel, and reading as many books as possible. Finding ways to combine all of these with ITAM & software licensing is always fun!
Connect with Rich on Twitter or LinkedIn.