This blog will look at an area of the business which might cause some people's eyes to automatically glaze-over, but my challenge is to take this potentially boring topic and flip it on its head. What am I talking about? The cost of cloud!
Cost seems to drive most conversations around cloud adoption, but we all tend to pretend it doesn't. Each cloud is different, everyone knows that.
But here's the news flash: the way that cloud providers charge for their clouds is equally as different, and it can actually be a dangerous conversation to enter if you're not equipped with the knowledge you need to navigate it well.
Let's start off with the number one issue, and it often goes right back to the point when your CEO first said: "Let's put everything in the cloud!" The migration phase. If you move your workloads 'as is' from your existing in-house data centre or broom cupboard in some cases to a utility type cloud then chances are this could lead to a large bill that looks more like a phone number; ten digits long.
Why? Mainly due to the fact the utility type cloud you are now on will charge you a per hour rate pricing model. Think of it as water out of a tap, electricity out of a wall or gas out of a pipe. If I gave you a laptop for free and said I will charge you every hour you have it on, I am sure you will turn it off whenever you can.
To efficiently use these clouds you need to re-think the way you will consume resources on them. This means more work on automation and diligent processes should be put in place to keep everything costing what you would expect. Sounds like hard work, but it can actually be quite cheap if you set the right processes in place and do it well.
Here are three quick questions you can ask yourself in order to know if your workload can run in this utility type pricing structure:
- Can my workload be turned off some of the time when it's not being used?
- Can my workload be setup in such a way it can increase and decrease in size when demand says its needs to?
- Can my workload run randomly over a given time period?
If the answer to any of those three questions is yes, then a utility type charging model may very well work for you. Taking this one step further; some super smart cloud providers out there have thought of ways to charge cheaper rates for unused capacity, or (and this is my personal favourite at the moment) paying for running compute code only!
Essentially, this means providers have said; "Why should you pay for an entire PC worth of cost when all you want to do is run some compute code?" This method charges you for the compute cycles used to run your code and it's measured in seconds, making it super-efficient! The cool kids at the milk bar call this "Serverless Computing", and it's the way of the future!
That's all well and good I hear you say, but what if my answers to the above questions were "no"? Don't worry, the good news is there are clouds out there for you too. Every cloud provider has a model of charging for these use-cases as well.
You pay the same amount whether it's on 24x7 or off for 23 hours in the day. It's best suited for traditional business applications which aren't ready for the big leagues. These types of clouds include on premise private which could span multiple racks, right though to full public models where you can purchase a chunk of preassigned resources.
As with most customers, at the end of the day you will probably end up with a mix of both, but after you've read this blog hopefully it makes a little more sense as to why. These two ways of thinking around costs can help you take the first step in planning out which type of cloud is for you.
The next time someone in your office says to you; "Which cloud is the cheapest to run my application on?" You can tell them to come back to you with a serverless design and then we can talk cost!
Keep across your portfolio with a single unified view.
For more cloud go to Telstra Cloud Services.