Introduction: Is cloud really a kind of magic?
Many have positioned cloud computing as the future of all IT. As part of this, advocates, evangelists and those with a commercial interest give the impression that cloud is universally cheaper, more flexible and delivers greater scalability.
For many years now, we at Freeform Dynamics have been advising business and IT professionals to be wary of this kind of generalised rhetoric. Sure, the right cloud service can be great when aligned with the right requirement and implemented responsibly. But with such a diverse range of offerings in the market, from myriad providers with varying degrees of robustness and maturity, it’s easy to get caught out.
Against this background, making decisions around cloud, particularly in relation to a service that your business will rely on, requires the same disciplined approach as you would apply to any other significant technology investment.
This principle came through very strongly in a recent research study in which 223 UK based IT professionals gave their views on technology-related decision-making in general. The main findings of the study are outlined in our report entitled “IT Decision-Making in the Digital Age”. The document you are reading now is a drill down looking at more detail of the views we gathered on cloud in particular.
A quick look at cloud-related activity
While our introductory comments above were intended to set a down-to-earth tone, this should not be taken as in any way dismissing the importance of cloud. Indeed, the first finding we shall look at from the research tells us that various forms of cloud service have already found their place in the enterprise IT landscape, with activity set to continue growing (Figure 1).
Click on chart to enlarge
Apart from illustrating the market momentum, this chart is a useful reminder that ‘cloud’ doesn’t translate to just one kind of service. In the crudest sense, we can align services with different layers in the traditional systems stack, from ‘Infrastructure as a Service’ (IaaS) at the bottom (which loosely maps onto the hardware layer), through ‘Platform as a Service’ (PaaS) in the middle (the middleware, database, utilities and tools layer), to ‘Software as a Service’ (SaaS) at the top (roughly equivalent to the application layer).
In this particular study, we also included reference to ‘Function as a Service’ (FaaS) in acknowledgment of the degree to which many providers are now making specific functions available via web service APIs (Application Programming Interfaces) for developers to embed in custom applications. Examples here include payment functions, mapping services, programmatic access to social networks, and even commercial transaction capability.
Of course even the above is a huge simplification of what’s on offer in the cloud space. Within each of the categories we have mentioned, you will find hundreds, sometimes thousands, of providers, each often delivering a portfolio of solutions. More of what that means a little later.
In the meantime, from an adoption perspective, the emphasis at the moment is at the top and bottom of the systems stack as we can see – i.e. SaaS at one end (convenient access to turnkey applications) and IaaS at the other (open access to compute and storage capacity). The frequently proprietary nature of PaaS and FaaS means they are a little behind at this point in time.
What’s driving and inhibiting cloud adoption?
When Freeform Dynamics started researching cloud 6 or 7 years ago, we quickly discovered the importance of avoiding too many assumptions on drivers and inhibitors. We found that if you ask questions in an open and objective manner, you invariably find that the same factor can be viewed as a positive by one person, while at the same time being seen as a negative by another. Furthermore, some of the supposed drivers and inhibitors frequently reported in the press are actually not that big a deal for many organisations. The results of our latest study are totally consistent with this (Figure 2).
Click on chart to enlarge
So why do perceptions vary to the degree we are seeing here? Well, the first factor on the list provides us with a good illustration.
Many services live up to the flexibility promise, allowing you to scale up, scale down, discontinue and so on, and only ever pay for what you use. Other services, however, work on quite a different basis; indeed, the words ‘rigid’ and ‘uncompromising’ spring to mind when you look at the terms of service frequently offered.
As a simple example, it is actually quite common for providers to lock customers into fixed term ‘ratchet-style’ contracts, allowing commitments to go up, but never down, and imposing costly penalties for early cancellation. And when the renewal date comes around, you are then forced to commit for a minimum of another year, making it very difficult to extract yourself from the arrangement at a convenient time.
Add technical constraints to this kind of commercial rigidity, particularly in relation to less mature services, and it’s easy to see why some say that a move to the cloud actually undermines flexibility. However, if you then consider that many organisations have fairly fixed and predictable requirements, the whole notion of flexibility in such cases isn’t that relevant, which explains the group in the middle.
It is beyond the scope of this document to go into this level of discussion for the other factors we see listed on Figure 2, but to give you a feel for what’s behind the variation in each, here are some quick high-level thoughts on each category.
We have already mentioned solution/system flexibility. Alongside this many highlight rapid adoption capability as a big cloud advantage, and this can mean a shortening of time to business value. For others, however, especially for more complex/critical applications, the impact is not that significant because most of the time and effort during a software or service implementation, for example, is concerned with requirements analysis, then application development, customisation, configuration and/or integration. And if you skip this upfront work, as cloud arguably encourages you to do, you may end up with lots of silos of capability that only partially meet business needs. You then need to go back and implement properly to unlock the full value, which can ultimately take longer.
Use of cloud services, particularly from one of the larger more established players, can often represent a great way of reducing risk. It’s hard to compete with the level of security and scalability offered by a competent provider with a hyper-scale, state-of-the-art datacentre. But if the location of that datacentre puts you out of compliance with data sovereignty regulations, or connectivity constraints threaten user response times, then those advantages are quickly undermined. From a security perspective, many make the point that large service providers in particular may implement all of the latest best practices, but represent a much more prominent target for cyber criminals, so the risk is ultimately greater. The reality, however, is that it’s impossible to generalise in this whole area – it depends on the nature of the provider and the service, and how what’s on offer compares to what you can achieve in your own environment.
Facilities and IT overhead
Moving applications and workloads out of your own environment into the cloud can clearly help from a facilities and utilities perspective if you are short of space in your datacentre, or finding it hard to keep up with energy related needs. However, when you bear in mind the investments you have made and any future expansion plans, and consider sunk costs, committed spend, volume-based pricing on energy, etc, it can sometimes make better economic sense to keep things running in-house. And when it comes to IT team convenience, while pushing a discrete workload into the cloud can be regarded as one less thing to worry about, if that now means managing security, access, compliance, quality of service, troubleshooting, support and other functions across boundaries, IT’s job could easily become harder.
Cost and commercial
Many argue that doing away with the need for capital investment and hardware/platform provisioning and configuration can reduce start-up costs and commitment from a budget perspective. Countering this are some of the contract commitment considerations mentioned above, and the fact that much of the implementation effort is still required anyway. Then, however, there is the simple fact that some cloud services are actually extremely expensive compared to on-premise alternatives when you look at the cost of ownership over an extended period. Again, it’s impossible to generalise here; it really depends on the service and how you are using it, but clearly some are already finding that long-term TCO can be an issue. Finally, depending on how you budget, the subscription payment model may actually be a disadvantage. Rightly or wrongly, many organisations still prefer a Capex-based approach with a preference for asset ownership.
The overriding message from all this is that it really is impossible to generalise when it comes to cloud drivers and inhibitors, so each decision needs to be approached and each option evaluated on a case-by-case basis.
The consequences of getting caught out
Unfortunately, some have been seduced by the ‘instant on’ nature of cloud services and haven’t necessarily conducted the level of due diligence they ought to have. In other cases, business people have led the charge to the cloud, not knowing what they don’t know, and inadvertently ending up with a service mismatch. Overriding all of this, we then have the genuine difficulty of dealing with what is still a relatively immature market, in which things are constantly changing and it’s often very difficult to compare like with like. Put all this together, and it’s understandable that a whole range of issues and pitfalls were highlighted by research participants (Figure 3).
Click on chart to enlarge
And if you think of signing up to a cloud service as being akin to getting married, as both parties continue to change over time, irreconcilable differences often arise which ultimately lead to divorce (Figure 4).
Click on chart to enlarge
Considering cloud in a broader perspective
The research actually covered a lot more than cloud, e.g. as part of the study we asked respondents to tell us how well they were delivering against key indicators on an IT performance scorecard. You can see more details of this in the section entitled ‘About the Research’, which is included at the end of this document. Suffice it to say for now that this allowed us to identify a group of high achievers that we labelled ‘Top IT Performers’, and studying the behaviour of these yielded some interesting insights.
Firstly, corroborating the notion that appropriate use of cloud services can create significant business value, our Top IT Performers are noticeably more active in this area. However, the data suggests that this is not at the expense of investing in the on- premise IT environment. Those exhibiting a higher level of achievement are also much more likely to be enhancing IT operations through increased automation (Figure 5).
Click on chart to enlarge
In addition, to driving both the cloud and automation agendas, Top IT Performers are more likely to be committed to other improvement initiatives. Activity here ranges from optimising IT service delivery through DevOps and continuous delivery, to driving greater efficiency within the datacentre in areas such as energy management.
The ultimate goal is sustainable performance
To finish off our discussion, let’s highlight a combination of a warning and an imperative. Firstly, while Top IT Performers stress the importance of continued investment in on-premise infrastructure, indications are that some ‘Others’ have taken their eye off the ball here (Figure 6).
Click on chart to enlarge
The picture we see here is arguably understandable. If you are having trouble delivering, or are under pressure to do more with less, it can be tempting to run to the cloud as a quick fix. The more experience you gain with cloud services, however, the more you realise it’s only part of the equation. The lesson here is therefore that when it comes to optimising service delivery in a sustainable manner, it’s important to acknowledge that the centre of gravity for IT will remain on-premise for the foreseeable future. As a result, investment in your datacentre remains critical.
Final thoughts and recommendations
While we have tried to provide a flavour of the kind of challenges that can arise with cloud services, our aim has not been to discourage you from going down this route, but to make sure that as you travel along your own cloud journey, you do so in full awareness of the things to look out for.
The question of if and when
Whether you should or shouldn’t be using the cloud is a question to be asked on a requirement-by-requirement basis. Bearing in mind all of the factors we have discussed, you will undoubtedly find that it makes absolute sense in some situations, but is a total non-starter in others. In between it will be a judgement call.
As you make decisions in this space, something to think about is the way things will change over time. Being able to rapidly spin up a few servers in the cloud for software development, testing and piloting purposes can be really useful, but the provider’s commercial model may make scaling up into production cost prohibitive. As a simple example, charges relating to data transport, DR/recovery arrangements, archiving and so on, can be an issue if the application is processing large volumes of data.
Another reason for pulling an application back in-house is the need to deal with evolving regulatory requirements and/or to make it easier to apply your corporate policies and controls in a coordinated and integrated manner. Of course this can work in the opposite direction too. If you have resisted using the cloud for a particular application because the provider you were considering did not have a datacentre in your region or country, if they later open a local facility, that may prompt a cloud migration.
Hybrid is the future
Looking forward, the real aim for the vast majority of organisations is not to shift everything into the public cloud and shut down your datacentres, but to have the freedom to run applications wherever is most appropriate – on-premise or hosted – and migrate freely as required over their lifetime.
The big requirement here is for what many refer to as a ‘hybrid-cloud’ architecture. It is beyond the scope of this document to go into detail here; suffice it to say that with the right platform software and tools, you can construct a private cloud environment that allows you to enjoy pretty much all of the flexibility, scalability and efficiency benefits of cloud, but on-premise. Solutions available today increasingly enable you to not just migrate things back and forth between public and private clouds with ease, but to apply policies and controls consistently as you do so, and maintain constant visibility of what’s running where so you can manage and optimise continuously.
The next-generation datacentre
Hybrid-cloud architecture is just one aspect of the next-generation datacentre. The same goals of holistic management, visibility and optimisation apply equally to the facilities side of the equation.
Minimisation of energy costs and consumption, for example, is key to achieving datacentre efficiency, and effective management of power distribution and related backup/fail-over measures within the facility are critical to minimising outages as a result of localised overload and other failures.
Modernisation of the datacentre power-infrastructure to create a more flexible environment is then important to get reconfiguration of the power supply off the critical path when installing, upgrading or moving IT equipment. This is a huge consideration as older servers and storage devices are replaced by the latest generation of kit that packs everything into the available space more densely.
Playing the flexible cloud game
If you zoom out and look at all this from a bigger-picture perspective, pretty much all of the technology, tricks and methods used by cloud service providers to create a robust, flexible, efficient and scalable environment are now available in an enterprise datacentre context. And the good news is that whether it’s workload management and migration, compute, storage and network optimisation, or taking energy efficiency to the next level, many of the latest solutions and practices are already proven by those running public clouds.
The upshot is that as time goes on and your own datacentre becomes naturally more cloud-like, the differential between the hosted and on-premise worlds will increasingly shrink. Does that mean that you will ultimately never have a need for public clouds? Almost certainly not – there will always be situations in which the right provider can deliver incremental benefits. The choice of public versus private cloud will become a lot less stark and persistent, however, making it easier to change and optimise things on an ongoing basis, and to correct mistakes should you miss something when conducting due diligence or if a provider lets you down.
The bottom line
The overriding message we would like to leave you with is that cloud is an extremely important development in the IT industry, but it isn’t some kind of magic – it’s just a different way of delivering IT services. The trick is applying the appropriate level of discipline and due diligence when making decisions, then use it where it makes sense, and avoid it where it doesn’t. In the meantime, make sure you continue to invest in your on-premise infrastructure as the chances are you are going to be needing it for the foreseeable future.