Public cloud has been a total game-changer for both businesses and individuals alike. Easy access, unlimited use, what’s not to like? Well, many business users of services like Azure, AWS, and Google Cloud are discovering one of the major downsides of public cloud – the unpredictable and unexpected costs.
It’s somewhat ironic. Cost is one of the main drivers for public cloud migration. People expect to save money because they’re not having to spend out on the hardware and infrastructure required to run a system of this size. However, those savings can be quickly lost through poor management, increasing use, data egress and more. Let’s take a general look at public cloud costs, and then get into those pitfalls that have so many businesses looking at alternative options.
What causes the unexpected costs of public cloud?
1. The pay-per-use public cloud model
Public cloud services like Azure, AWS, and Google Cloud charge on a pay-per-use basis. For organisations that have found themselves paying for server capacity they are not using the pay-per-use model is attractive. And, as companies look at their options when their hardware is reaching the end of its life, the relatively low monthly cost of public cloud looks particularly appealing compared to the expense of new infrastructure. The pay-per-use structure will be cost-saving for a lot of businesses.
However, pay-per-use does not necessarily equate to cost-effective, especially in the long-term. Hotels are technically also pay-per-use, but for most of us, you can’t afford to live in them.
An unplanned increase in use of the cloud can easily lead to your bill at the end of the month being higher than excepted, and this can happen in a range of ways, including but not limited to:
- A workload left on for longer than necessary
- A surge in data egress and the fees attached to this
- An unplanned increase in workloads
And these situations will only become more common as your cloud grows, more data is stored on the cloud and the number of workloads increases.
2. The add-ons
As well as your base public cloud system, there's a good chance you'll be paying additional, optional, charges, which might not have been included in your initial cost-estimate for your public cloud system. While I use the word 'optional' and it is technically correct, a lot of these add-ons are actually essential to your business, and can't be avoided.
- A range of cyber security solutions
- Support services
- Performance and accessibility improvements
- Ongoing migration costs
- Extra fees for hosting your cloud in the specific country/region you need for data compliance reasons
These add-ons can add up to your final bill being a lot higher than what was originally budgeted for when you made the decisions to launch this public cloud, especially as time goes on, your needs change, and more add-ons are needed to keep up.
It can also be really difficult to keep track of exactly what you're paying for since cloud bills are not often broken down in a business-friendly way. The granular nature of a cloud bill doesn't explain which charges are related to which workload, but instead they're granular and break down into things like costs for balancing, data storage and transfer charged per GB, other services charged by minutes of use, VPN charges, egress charges, etc. So it's understandable to lose track of exactly what you're paying for, and end up paying too much.
How to reduce these unexpected cloud costs
Whether you are an Azure, AWS, or Google Cloud user, the risk of unexpected and unpredictable costs is the same. It’s not like these companies are laying traps for you to fall into, to trigger these unexpected charges. In fact, they often offer tools and support to avoid just such a thing. Most likely, high costs will arise from misguided expectations, poor management, or simple mistakes – in other words, user error. Knowing what causes these unexpected costs allows you to avoid them.
1. Avoid false expectations
It is unrealistic to expect you can cap costs simply by controlling cloud use. The reason for this is simple – most people don’t understand how cloud costs work, so the only way to control this is to educate your workforce and put in incentive programmes to encourage them to limit their cloud use and hopefully eliminate unnecessary costs.
For example, Google Cloud user Spotify, have introduced cost management schemes using visual dashboards to help remind and encourage staff to reduce cloud spend. Azure, Google Cloud, and AWS also have built-in tools available to help you control your spending.
You can't view this as a top-down issue, since it must be a company wide directive to reduce cloud usage. ‘We must reduce cloud spend’, without any initiatives to help your teams understand how costs are building up will yield very little result – just more confusion. At the same time, having these company-wide conversations will highlight when cloud use is unavoidably increasing.
2. Don't put the wrong workload on the wrong platform
Public cloud works better for some workloads than others. For example, for customer-facing services where ease of use and access is the priority – like Spotify, Netflix and the like – public cloud makes sense. However, if you are managing very large data bases with a lot of reads and writes (i.e. a lot of use), costs can mount quickly. Likewise, if you are working with very large files that people prefer to view online but edit offline – such as videos or CAD files – the process of downloading those files (data egress) will be both costly and time-consuming.
If these data egress costs and other data managements costs are getting out of control, you need to consider if a change of process/policy around this data could reduce costs by reducing interaction with the cloud. You also need to consider if part of your IT needs to be stored on an alternative infrastructure platform.
3. Don't make expensive mistakes
Some workloads run 24/7, but others are business hours only, or even sporadic – perhaps for just a few hours, or in a kind of ‘binge’ run every so often (for example: when there's a billing cycle and your account department need to process all of the transactions in one afternoon). Virtual machines are charged by the second, so for those workloads that can be turned off, it’s vitally important to remember to do so. Otherwise, you are being charged for compute capacity that you are not using – and wasting money. This is one of the trickier cost-reduction policies to follow, since it's all about not forgetting about these workloads.
Azure, Google Cloud, and AWS offer a scheduler tool that enables you to control start and stop times for VM use, which will help in those instances where VMs can be regularly scheduled and turned off at the end of the day. For more sporadic workloads, though, a scheduler won’t help – and again it comes down to instilling the right behaviours in your team to ensure this doesn’t happen. Whether through good old fashioned education, some kind of incentive, or even a penalty, you need to make sure that everyone appreciates the cost of this kind of mistake, which could quickly run into the thousands of pounds.
What if you can't get these costs under control?
All of the above cost-saving advice have the same unfortunate things in common, they need commitment from the entire business, and they require careful controls over the use of public cloud. And it's very easy for these controls to become restrictions, interfering with your employee's ability to work.
If you're struggling to get these unexpected costs under control without impacting the productivity of your cloud users, then it's time to look at alternatives. Migrating to privately-held infrastructure, whether that is private cloud or your own hardware in a colocation centre, gives you greater control over costs in the short-term, as well as the long-term cost benefits of ‘owning’ your system compared to ‘renting’. And with the right provider, you don’t need to miss out on the ease-of-access and use that public cloud offers.
Better budgeting ahead
The first step in getting unexpected public cloud costs under control is making them expected costs. Identify where your big spending is happening, then you can start to wind down the unnecessary costs. Working with your employees to reduce unnecessary public cloud use will yield significant savings, and pay many times over for any incentive programmes put in place.
But if you're finding that you can't reduce your public cloud use and the costs are out of control, then you might have outgrown your current cloud platform. Moving to an alternative can be more cost-effective and provide more consistent and predictable bills. Private cloud and colocation will give you greater control over your spending, and a hybrid IT solution will let you retain your current public cloud while moving some workloads to more cost-effective platforms.
If you want to learn more, get in touch with our expert team.