A clear outlook for cloud computing

I think it's fair to say that the cloud has become ubiquitous with IT. If you are building a new platform the chances are you will build it in the cloud.? If you have an existing platform, the chances are you will migrate it to the cloud, or at the very least replace it with a cloud hosted alternative.

And it's not surprising that cloud technologies are often cited as being cheaper, faster and safer.

However, recently there have been articles stating the opposite - that moving FROM cloud into private data centres has saved companies money and made them more secure - a process known as Cloud repatriation.

So what's going on? Have we been lied to? Are we seeing a modern day enactment of Hans Christian Andersen's 'The Emperor's New Clothes?' Is some young child about to shout - 'Cloud is rubbish!'?

First, you need to fully understand why, for some use-cases, cloud might not be the silver bullet everyone thinks it is. In the case studies I have come across, the systems where money was saved moving FROM cloud have followed a specific pattern - that being a system that is already big, but also with relatively slow and predictable growth, with little to no variance in usage.

A typical example is a platform that is largely built around storage - because storage use rarely decreases, and you can more easily forecast how often that storage is filled and therefore plan in advance adding additional storage as needed over a longer period.

However, how many services are like this?? Although all services use storage, even a lot of storage, the real value of all that data is actually in the processing of it - but once it's been processes you no longer need the compute, hence for most platforms the ability to scale back on that compute - which is usually the most expensive - can save you huge amounts in the long run.

Which brings us to a common reason why companies may become disillusioned with cloud - they migrate traditional Virtual Machines running monolithic applications and leave those VMs running 24x7, then wonder why they aren’t saving money.

In fact I wrote a blog post? that compared traditional style usage against re-building a solution to use more dynamic services: Are you barking up the wrong tree with your Cloud Strategy? (ir77.co.uk).

I won't re-iterate everything in that post, but I'd encourage you to check it out.? The short version is this - if you have a VM running 24x7, it could cost you up to three times more than a serverless solution that followed a 9-5, weekday usage pattern.

As newer technologies such as AI come to the fore, this flexibility around computers becomes ever more important.? Specialist AI chips that power the latest Gen AI capabilities cost $40 000 to buy, whereas to use Azure's Open AI service costs nothing while idle, and as little as $0.01 per individual query - so that means you'd need to use it a LOT to get $40k worth.

Finally, one area that is often overlooked is the ability to quickly create environments, try out a new technology, then destroy it - all within a few hours and at low cost.? Again, this becomes evident with newer AI technologies - as per the previous example, you can spin up an Open AI instance, play about with it, then shut it down for less than the price of a coffee.

So, although I agree cloud technologies aren't right for everyone, more often than not, the problem is when the technology is used in the wrong way that you can experience increased costs - and this can be fixed by re-platforming or re-architecting your solution.

If you'd like to help migrating to cloud, or would like a review of your current IT landscape, get in touch and see if Iridium can help you reduce your cloud costs whilst increasing security, compliance, performance and scalability.