We’re about being there for our clients when it really matters. When it absolutely has to be right. Doing the right thing is both our ethos and sweet spot. And it’s why clients turn to us again and again
Find out how you can make a big change to your career by joining one of the best small firms in the UK.
I graduated from the Helsinki University of Technology with a MSc in Industrial Management and after a year travelling around the world, joined Accenture.
Shortly after starting my career I met m...
Kare Heikkila, Consultant
View Kare now
It's the people that make Berkeley different to other consultancies.
Bright, friendly, down-to-earth people who are both thinkers and doers. Working by your side, as consultants and colleagues, to get the right results.
Our Partners and consultants share their perspectives and thinking on topical issues.
Home > News & Views > Articles > Moving your data centre to the Cloud
Most companies have already been through a first phase of driving cost efficiency in their IT infrastructure, taking measures such as consolidating into fewer (or one) global data centres, and outsourcing infrastructure operations. However – according to the hype, at least – The Cloud (or infrastructure as a service, IaaS) promises to offer a whole new wave of cost reductions, efficiency savings, and service improvements, but probe just a little deeper, and some big questions quickly become apparent:
We have recently been working with two clients – one a FTSE 250 manufacturer, one a FTSE 250 consumer goods company, both global brands and household names – on pioneering programmes to move their data centres to the Cloud. Based on that experience, this paper explores some practical answers to those big questions.
Cloud is one of those terms, like Big Data, that has become so frequently used and hyped that you will see the terminology used – whether justifiably or not – to describe a very wide range of services. For the sake of IT Infrastructure as a Service (data centre services), we use the model and terminology as shown in the table below.
So the first key point to note is that when you go out to market to procure Cloud services, you may get a wide range of responses in any one of those 4 categories described below, or they may have characteristics that span more than one category. Typically, the ‘industrial strength’ offerings that are suitable for big, mission critical, transactional applications are the middle two – with Virtual Private Cloud arguably offering the most promising ‘best of both worlds’ combination, many of the cost and agility benefits of public cloud, but with the control, tailoring, security and risk-mitigation benefits of traditional hosting. We don’t currently see public cloud (most famously Amazon Web Services) as being the preferred option for these sort of industrial strength corporate applications – other corporate uses certainly, but ready to host the ‘big iron’ transactional ERP systems of some of the world’s largest companies? Probably not yet – but it’ll be interesting to see how the space develops. We’ve helped one of the world’s largest energy companies move the entirety of their external web site hosting to the public cloud and we are also helping one major automotive manufacturer put their SAP CRM instance on public cloud (with a 3rd party service wrapper). But we haven’t yet seen anyone put true, ‘backbone’ transactional ERP on public cloud.
In determining which of these models is best for you the key, as ever, is to understand what your needs are, i.e. what characteristics of the service really matter for you? These characteristics tend to fall into two groups:
• Utility pricing – A Cloud service should really offer utility pricing, i.e. a true pay-as-you-go service, with zero fixed costs, which can be scaled up and down (by compute power, by storage volume, by number and type of Virtual Machine, etc) as much as you like – which is clearly the most beneficial model in terms of flexibility.
• Asset ownership and refresh – a pure Cloud service should mean you don’t own any of the IT hardware assets; and nor should you pay for their refresh & replacement.
• Legal contract (terms and conditions, Ts&Cs) – this is an area often overlooked by people more interested in the technology, but the legal / contractual aspects of committing to a Cloud service are actually an important area of differentiation.
• Scope – the scope of a typical Cloud service in what respect is usually IT infrastructure (computer, storage, networks), including all the necessary backup, service resilience, and disaster recovery contingencies; and software support only up to the OS, database and SAP Basis layers of the stack.
• Virtualisation and multi-tenancy – a pure Cloud service will use virtual machine (VM) technology and shared storage such that none of the physical hardware will be specific to you; it will be shared with other customers of the service, like multiple people sharing the same house (hence the term ‘multi-tenanted’). But clearly, in this case, you need to ensure that your data and compute power (your ‘room within the house’) is sufficiently secure and isolated from any potential impact of other tenants.
• Service model – for a pure Cloud offering as for the technology, much of the service operations will be a shared service as well. But again, many of the Virtual Private Cloud and Private Cloud offerings will enable you to tailor this – such that you have bespoke service levels, or bespoke support arrangements (for example, you might want a dedicated, in-country, local-language-speaking, service manager in certain key locations across Americas, Europe and Asia).
• Data sovereignty – where your data is physically stored, and therefore what legal and regulatory regime applies in terms of data protection and management, is another key consideration. Perversely, the United States – highly fiscally and politically stable, and with world-leading technology skills – may be one of the least attractive countries due to the implications of their Patriot Act legislation.
This may sound like both having your cake and eating it, and we were originally sceptical that both could be achieved. But the reality is that for a well run and well implemented Cloud service, the pure economies of scale mean that it should be able to offer a better service for a lower cost. But obviously, the big caveat here is that it all depends on your starting point (in terms of how cost effective, and how good a service, your previous IT infrastructure operational arrangements offered).
In our practical experience of migrating the entirety of a global datacentre for a FTSE 250 manufacturer (both SAP and all non-SAP, Windows and Unix), the transition will have achieved:
• Total operational cost savings (on a cash basis) of 42%
• Service level improvements – most key SLA metrics (for availability, incident & change management, etc) were better than previously, and the remainder were the same (i.e. none were worse, and the net position was certainly an improvement)
• Flexibility – a largely (but not completely) utility-based pricing approach, with a 50% capacity floor, means there is ample room for major service changes (e.g. to support corporate acquisitions and divestments) with the commercial model flexing on demand, and little risk of being left over-or under-capacity
• Responsiveness – the time to stand up new environments (e.g. new development and test environments for projects) will go from a matter of weeks to a matter of days (albeit still not hours / seconds like public cloud would be; but that is simply not a requirement for this particular client)
• Security – having done extensive review and analysis, the client’s internal IT security experts considered the Cloud service to offer the same or better IT security overall than the provisions under their previous hosting arrangements – despite the multi-tenanted approach. Importantly, the contractual recourse (i.e. warranties, indemnities and limits of liability) in case of security breaches is also better than their previous arrangements.
The suppliers of these types of Non-Public Cloud Services generally fall into three groups, each of which has strengths and weaknesses as shown below (although we risk some unfair generalisation in this list – clearly, each specific supplier will have their own strengths and weaknesses):
Again, it’s worth considering both the commercial risks and also the technology & service risks. Although this is by no means an exhaustive list, some of the common themes(in our experience) of risks and challenges we see across clients in this area are as follows:
A year ago, our answer would have been ‘maybe’. But as of now, our view would be an unequivocal (although still cautious) ‘yes’. There are now a number of case studies of where this has been done at large scale, for major global companies, and for their most mission-critical transactional applications – so it is still advanced, certainly, but no longer ‘bleeding edge’. And the benefits of very material operational cost savings (approaching 50%) and gaining some worthwhile service level improvements are simply too good to ignore. So our view would be unquestionably that if CIO’s are not exploring this already, then they should be – and will be soon.
Download this article
Creating a corporate strategy from within the business
Identifying opportunities for collaboration between Operating companies
There is lots of hype around Cloud, and confusion between public and private Clouds and what they can offer. But navigate that successfully, and there are some genuine and exciting benefits to be realised by migrating data centre operations to a Cloud-based approach.
Jon BradburyPartnerContact Jon Bradbury