Why multi-cloud is the New Normal

Not long ago the IT landscape of an enterprise was a simple affair: one or a few data centres, usually determined by location requirements, would create a system landscape that every IT person with 10 or more year’s work experience grew up with. Fast forward to the 2010s and the landscape was characterised by duality – meaning that the data centres were still there – and a few workloads in the cloud, most likely a single cloud. Take another jump to present times and the landscape has fundamentally changed: some (typically the smaller) data centres are gone, the existing ones usually have a smaller footprint and between one and two thirds of enterprise workloads run in the cloud. Not one cloud – but many clouds.

The consequence is an order of magnitude of more complex operating infrastructure for enterprises – and all that happened in the run of the last 15 years. So how did we get here? 


IT took a wild ride – or was it ridden wildly?

15 years ago, life in IT was as good as a life in IT could get. There was of course cost pressure that had led to virtualisation, but it was almost all on VMware, so IT enjoyed a single tool that literally ran all enterprise workloads. The fabled single pane of glass of managing all enterprise workloads was reality. 


Phase 1 – Till 2012 – IT in charge

Then the first cloud projects happened. Typically, this was with the enterprise choosing an early cloud-based vendor (e.g. Salesforce or Workday, two early cloud vendors that are still running strong today). These early SaaS vendors had to run their offerings in their own clouds, so apart from having integration needs that went outside of an enterprise’s data centre, they did not add much complexity for IT. After all, CRM and HCM products had to be integrated in the on-premise world as well. That changed when enterprises had to go into the cloud for the first time themselves, usually with demands to build custom software (e.g. for a digital transformation project, a BigData initiative, an IoT project etc.). At this point, IT would choose its ‘preferred’ cloud partner in the realistic hope that the enterprise would move forward with this one cloud vendor in the future. After all, IT had managed to establish a single vendor strategy in many of the IT categories. 


Phase 2 – Till 2018 – IT is ridden by the business

And then – all hell broke loose for IT. Analysts propagated that CxOs would spend more on IT than the CIO (the CMO being the most prominent). The idea was flawed from the beginning, as CIOs were always spending other CxOs budgets – to implement solutions for them. But at the core was an all IT changing truth: CxOs could choose their application vendors and did not have to work with IT on that – at all. A stark departure from the past, and well, CxOs did take advantage of that. Suddenly it only mattered that a solution ran in ‘the cloud’ not in THE cloud that the CIOs wanted the enterprise to standardize on. 

Things were not better on the custom project side which ran the digital transformation initiatives of the enterprise. Several trusted IT partners’ cloud offerings failed (Cisco, Dell, HP, VMware – just to name a few) and any enterprise that had picked them for their workloads had to move off them. In other cases, some clouds turned out to be less strong in an important category – and enterprises found themselves running their digital transformation projects in multiple clouds. Consultants from SIs would often recommend different clouds. As it was characteristic for the early stages of digital transformation – speed was of the essence – the clean up would come later. 

The resulting IT reality was a complete mess: integration had to happen across on-premise data centres and various clouds. The same data was often uploaded to the same cloud multiple times, so frantic was the pace and demand on integration needs. IT lost its single pane of glass and in many cases, divisions would opt out to their SaaS vendors for support and monitoring of their workloads. 


Several trusted IT partners’ cloud offerings failed – Cisco, Dell, HP and VMware just to name a few


Phase 3 – From 2018

When things could not get any worse, they changed for the better, mainly due to two major trends:

  1. Firstly, Kubernetes established itself as the new container platform that was adopted across the vendor landscape. This allowed IT to get its hope back for the single pane of glass to run enterprise workloads – no matter where they run. 
  2. Secondly, the cloud stacks have come to the local data centre. AWS, Google Cloud and Microsoft Azure (in alphabetical order) all allow for the operation of parts or all their application technology stack – on-premise. The benefit for IT is enormous, as the consolidation of technology stacks reduces IT complexity substantially.  


New IT Strategies for the new normal: Multi-Cloud


The new reality requires new IT strategies and they are characterised as follows:

  • Consolidation is a must. Complexity reduction is of paramount importance. Bring data centres together. Bring databases together. Standardise on integration tools. Centralise security. Implement a single MDM tool. Standardise end point controls. Use suites as much as possible and bring – in the ideal case – the number of strategic infrastructure vendors down to five (ideally) or seven (if you must). 
  • Standardise on Kubernetes. To achieve the single control pane again, IT needs to standardise on Kubernetes which is the only technology that by now has a good shot at returning this to IT. It is critical that new SaaS and PaaS workloads are available on Kubernetes to avoid further fragmentation again. The good news – pretty much everybody in software is moving to Kubernetes as it has (for now) won the container wars. 
  • Embrace multi-cloud. There are upsides to operating in the multi-cloud. Not only from a negotiation standpoint but also from an uptime perspective. Though cloud outages are rare, they can happen, and if high availability is critical for some workloads, only the multi-cloud offers enterprises a reasonable shot at achieving that state of uptime. 
  • Fight for the single pane of glass. With acceleration being paramount for enterprises, IT cannot operate with multiple dashboards. To reduce complexity, IT leaders must do all in their power to get back to the single pane of glass. Beyond that, they need to push for automation solutions which are not just human operated. The current hype around ‘visibility’ is good only to a point, that routine IT operations in 2021 and beyond – need to be operated by software – not humans. 


So, there you have it. Did IT take the wild ride – or was IT taken for a wild ride? You decide on your own. The key takeaway is: good IT strategies were always crucial to enterprise success – and that has not changed in the 2020s. With the dominance and industry consolidation around Kubernetes, there is a good shot for enterprises and IT to get back in control of their workloads, remarkably in a much more complex system landscape, which is the multi-cloud. Multi-cloud is the new reality and the new normal – probably for at least a decade to come, likely longer.