Given how prevalent it is, you might be forgiven for thinking that everything is in the cloud today. But according to the latest research from the Cloud Industry Forum (CIF), the average amount of IT infrastructure hosted in the cloud by any one company is currently hovering around 50%. In fact, only 7% of organizations currently have more than 80% of their infrastructure in the cloud – but at the other end of the spectrum, only 8% of businesses have less than 20% of their IT in the cloud.

In short, the cloud is mature, but nowhere near universal.

And when it comes to future adoption, almost half (42%) of businesses have a ‘cloud first’ policy, with 55% taking a hybrid approach. Only a small number of organizations (2%) currently keep IT on-premises wherever possible, so there’s ample evidence that cloud is well and truly a proven technology today.

Furthermore, many organizations are now using more than one cloud, fitting the cloud to the workload, rather than the other way around. We’ll look at what this means a bit later, but first, let’s look at some of the common benefits and challenges of migrating from on-premises to cloud.

Cloud’s mature era: The benefits

The main benefits of cloud are well-understood. According to CIF, IT professionals adopt cloud to improve agility (32%), increase ease of scaling (29%) and gain cost savings (29%). One of cloud’s main benefits is that it can be tested, spun up and turned off at the drop of a hat, allowing organizations to only use the resources they need, when they need them.

Furthermore, a whopping 81% of organizations believe that cloud makes their IT estate either significantly or slightly more mature than using purely on-premises infrastructure. This is largely for two reasons: firstly, most cloud providers offer security at scale. They invest in the greatest and most resilient security with both physical and digital security constantly being tested and evolved. Secondly, although a larger IT estate means a higher number of endpoints to secure, it also means a more dispersed estate, so any one outage is less likely to have a great impact.

However, migrating to the cloud is a difficult journey. In fact, half of IT decision-makers say that migration was more complex than expected. Let’s look at some of these challenges.

The challenges of moving to cloud

The ease of moving to cloud is a bit like moving house: it depends what you’re moving. Leaving home to move into your first house is a lot easier than moving a family of four after a decade in the same place. That said, there are a number of pitfalls common to most migrations.

Firstly, moving to the cloud takes a significant amount of due diligence, which makes choice of technology partner important. Evaluating the most suitable workloads and applications to migrate to cloud, taking into account technology lifecycles and organizational needs, is a complex and difficult piece of work.

Companies also need to consider the likes of resilience, backup and disaster recovery, as well as network, cybersecurity, and automation requirements – all in advance of even choosing a cloud provider. With the complexity of cloud migration, many end-user IT teams come to realize that there’s rarely a single cloud that will cover all their needs.

There’s also the longer-term work to consider – for example, setting the vision and strategy, governance, establishing the roadmap, managing adoption – including training and feedback – as well as auditing success and looking at the renewal process later on.

Handling this complexity is probably one of the most difficult parts of moving to cloud. Every organization has a different technology estate with its own restrictions, policies and stakeholder preferences. Beyond this, businesses in more regulated sectors like healthcare or the public sector will have guidance and due process about cloud usage, including what can be stored or processed where and by whom. This makes the initial preparation and profiling of the business doubly crucial.

And as we’ve said, cloud can often make an organisation more secure, but it’s important to be aware of the security nuance of outsourced servers. For example, using bare metal will allow an organisation complete control over security options, but also the responsibility for doing so. On the other hand, public cloud often has parts of the security pre-set by the vendor, which can mean less power, but also less responsibility.

Finally, without meaning to compound the issue, most organizations today use more than one cloud.

The multi-cloud era

According to our own research, 80% of UK organizations either use more than one cloud today or are exploring or transitioning to a multi-cloud environment – and 95% of businesses believe that their use of multi-cloud will either stay the same or increase over the next five years.

In many ways, this makes a lot of sense. Different clouds suit different purposes and workloads. However, as with cloud migration, there are a few common areas to look out for when choosing your next cloud – particularly if you have multi-cloud in mind:

– Price: Your cloud needs to provide the right performance at the right price – but beware of vendors offering high initial discounts. These can creep down as the years go by, leaving you with a lot of infrastructure in a cloud environment which is becoming more and more expensive.

– Security: Know your data. If your data needs to be in a SecNumCloud environment, for example, find a provider which has operated one for years.

– Interoperability: If you’re running connected multi-clouds, they need to operate well with each other. If possible, look at testimonials from other users who are doing similar things to you before you move, and consider using open-source technologies to ease this and any future moves.

– Geographic reach: If all goes well, will your project expand to other regions in future, and if so, does your chosen cloud provider offer these regions? – Availability: When does your project need to start, and how soon can your chosen provider actually deploy the kit? If there’s going to be a lag, that’ll delay the project.

AI

No article today would be complete without a mention of AI. Specifically, it’s worth noting that sometimes your ‘primary’ cloud provider might not also offer cost-effective GPUs, so if anything, AI is fueling the growth of multi-cloud. Again, the decision about choosing a provider for AI workloads is nuanced: for model training, high-spec GPUs are usually needed, but because the data isn’t actually in production, latency isn’t an issue – so you can deploy anywhere. However, during production, latency (and therefore location) will become more important, but you can usually use lower-spec GPUs after the intensive model training is done.

Once a cloud migration is successfully completed and the pitfalls navigated, the journey is rarely over. However, this can be a very good thing: moving to cloud can help organizations to later explore the benefits of cloud native technologies. AI, machine learning and harnessing the power of big data can offer significant potential to a newly cloud-native organization – so if stakeholders aren’t excited by the simple migration to cloud, it’s worth looking two steps ahead and thinking about the future just around the corner.

We’ve featured the best cloud storage.

This article was produced as part of TechRadarPro’s Expert Insights channel where we feature the best and brightest minds in the technology industry today. The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/news/submit-your-story-to-techradar-pro