Cloud and data centre trends roundup 2016: Machine learning, hybrid cloud and Google's enterprise ambitions

A decade on from the launch of Amazon Web Services (AWS), the cloud market is continuing to evolve quickly. What was once seen as a toy for test and development purposes now hosts mission-critical workloads for some of the largest companies in the world, while vendors work on the next generation of cloud services, such as those around machine learning.

Business demand clearly shows no sign of abating. Gartner claimed the overall cloud market was valued at $208.6 billion in 2016, amounting to a 17.2 percent increase from $178 billion the year before.

As 2016 draws to a close, Computerworld UK takes a look back at some of the key emerging trends in cloud computing this year.

Google emerges as real cloud contender

There is little doubt that AWS remains the clear leader in the public cloud market. However, over the past 12 months competition has become fiercer.

Google has been offering a range of public cloud services for some time now, but during 2016 it became clear that it is taking the enterprise market a whole lot more seriously.

On the back of hiring former-VMware CEO Diane Greene to lead its enterprise cloud division, the company rolled out a number of high profile business customers at Cloud Next in San Francisco earlier this year, including Coca-Cola, Disney, Home Depot and Spotify, which migrated some systems away from AWS’s cloud.

Read next: Google takes aim at AWS - now it must convince CIOs it is right cloud for the enterprise

There was a concerted effort to appeal not just to the developers that Google has held sway with in the past, but also with CIOs and IT managers. It has also committed to a global data centre expansion to allay corporate data sovereignty concerns, with a data centre planned for the UK next year.

It is not the only firm threatening the dominance of AWS. Not so long ago it appeared that there was no catching the cloud vendor, with Gartner claiming it has 14 times the capacity of its competitors combined.

Traditional IT giants HPE and Dell may have already thrown in the towel, but these days there are a growing number of public cloud options for UK customers, with Oracle growing its IaaS business since launching earlier this year (Q2 results indicated $175m in revenues), alongside developer-focused Digital Ocean, and the cloud division of Chinese e-commerce giant Alibaba, which recently opened data centres in Europe.

A new buzzword arrives: multi-cloud

Increased choice has resulted in enterprises procuring services from more than one provider, picking a different cloud for a different business need. According to the RightScale 2016 State of the Cloud report, 82 percent of businesses already procure cloud services from a range of sources.

As the name suggests, multi-cloud involves reducing reliance on just one vendor, which can lead to benefits such as cost reduction. Analysts at 451 Research claim that enterprises can cut direct cloud expenditure by up to 74 percent with a multi-cloud approach.

There are also advantages around risk mitigation, while diversification helps avoid a new generation of lock-in in the cloud.

Of course, there are challenges in managing more than cloud provider. Moving applications between providers is not a simple task either, particularly for legacy applications.

According to 451 Research, the reliance on multiple cloud vendors will open doors for systems integrators and cloud brokers to help businesses to find the right provider.

Hybrid cloud gets boost with AWS and VMware partnership

One of the biggest cloud announcements in 2016 was the partnership between AWS and VMware. It was an acknowledgment that, for most enterprise businesses, moving all of their infrastructure to the cloud is unlikely to be a reality in the short or medium term.

For AWS, which has previously dismissed the “archaic” notion of on-premise private clouds, the deal enables it to tap into VMware’s unparalleled presence in corporate data centrs. This provides it with hybrid cloud capabilities that it has lacked thus far - an area that main competitor Microsoft already excels in, and will expand on with the release of its Azure Stack software next year.

Hybrid cloud has been discussed for years, but the concept is now being taken a lot more seriously with the partnership between AWS and VMware and the realisation that on-premise data centres have life left in them yet.

Cloud giants woo developers with machine learning and serverless computing

During 2016 it was clear that the big cloud providers are switching their attentions from commodity cloud server and storage sales to more lucrative services such as machine learning.

The big cloud providers released new APIs this year that allow developers to tap into machine learning expertise. AWS, Microsoft, IBM and Google all provide services around natural language processing and image recognition. There are also open source tools such as Google’s TensorFlow.

At the same time, AI personal assistants have gained mainstream attention - even if not in a positive sense, in the case of Microsoft’s Tay bot - and more and more businesses are investigating how to use the technology. However, there are lots of other interesting applications for machine learning, and open source libraries such as TensorFlow are enabling companies like Ocado to create more efficiencies in their operations. Read next: Ocado to replace barcode scanning with AI 'vision' to speed packing processes

There has also been a focus on the underlying infrastructure to supercharge machine learning, with the leading vendors offering cloud customers access to GPUs and FPGAs targeted at AI use cases, while Google announced that it is creating its own TPUs [Tensor processing units]. In fact this is where Google - which is currently a distant third behind AWS and Azure - hopes to gain ground in the cloud market.

Aside from machine learning, serverless computing has emerged as an increasingly popular way to build applications without worrying about the underlying infrastructure or configuring VMs. AWS’ Lambda has proved popular, but Google Cloud Functions and Microsoft’s Azure Functions have provided alternatives.

Read next: What is serverless computing and how is AWS Lambda helping enterprises run it?

Hyperconverged infrastructure

While there were interesting developments in the public cloud, increased automation in data centre infrastructure has offered improvements for those wanting to keep data on site.

Read next:Hyper-converged infrastructure: How software is transforming the data centre

Hyperconverged infrastructure has been around for some time - but the technology got a real stamp of approval with the September IPO of Nutanix, which essentially created the market segment back in 2011 with its first product launch.

There has been consolidation in the market over the past few years, not to mention companies going bust, and this could well continue with the traditional hardware-sellers taking hyperconverged very seriously.

Rumours have been flying around that HPE could be set to snap up Simplivity to boost its own position. However, at the same time HPE has been busy stirring up interest in its Synergy composable infrastructure platform, which it bills as the next step on from hyperconverged, essentially allowing its software-defined infrastructure to be set up and torn down quickly to support any application, whether legacy or cloud-native.

Read next: Hyperconverged infrastructure vendors: The main players compared


Copyright © 2016 IDG Communications, Inc.

Shop Tech Products at Amazon