As cloud computing becomes more pervasive, the cost seems to be tumbling. This is great news to organisation taking their first steps into using public cloud services – or is it?
In many ways, yes it is. Hardware costs are coming down, and modern tools are enabling more cloud functions to be automated, eliminating costs in physically managing the cloud with error-prone humans.
Software to layer on top of the infrastructure is either open source, where only support costs need to be covered, or commercial, where suppliers are finally figuring out that new, more flexible pricing models are required.
So, as the cost of the platform itself falls, it is only to be expected that a massively competitive market, such as the cloud, passes on some of those savings to its prospective customers.
However, this race to be seen as the provider of cheapest cloud services has its perils. Some cloud providers do not have the economies of scale to be able to buy hardware or tools at the same price as a much larger competitor. Therefore, to be cost competitive, savings have to come from other places.
Some of these may be valid. For example, using a datacentre facility positioned in a low-cost region can result in massive savings compared to a datacentre that is in a major financial district.
Effective architecting of the datacentre to minimise energy requirements, such as through the use of free air cooling, can lower energy costs by 40% or more. A small cloud provider may choose to use a co-location partner, so gaining access to the economies of scale of the facility’s owner, so controlling costs.
More on datacentres
The problem comes when the cost savings are created through cutting corners. Some cloud providers can only be cost competitive through not building in things like high-availability equipment redundancy – an “N+M” approach of having more hardware than needed deployed in a manner that any failure of one or more components of the platform will not result in a failure of the total platform.
N+M server cluster configuration refers to approach in cases where a single cluster is managing many services, and having only one dedicated failover node may not offer sufficient redundancy.
While all the equipment is running, customers will not see any problems – however, when a critical component does fail, access to the given cloud service fails too, and the impact on the customer’s business could be serious.
In such cases, the cloud provider should not be blamed for everything. Sure, they should have ensured that such failure didn’t happen. However, it is far more the fault of the customer that did not carry out due diligence on the provider and simply picked one that was the cheapest than the one that was more fit for purpose.
And this is the crux of the matter. Cost should never be the main reason for going for anything: it has to be whether the resulting environment supports the desired outcomes. IT has an unenviable track record for poor memory: many outsourcing projects have failed due to cost being the major concern, rather than capability. Many major projects have failed where initial cost was seen as the main concern, rather than flexibility and on-going costs.
Cloud contracts seems to be joining these as many organisations compare costs from different providers and just go for the standard “this one is cheaper” option.
When negotiating a cloud contract, by all means aim for cost predictability
The imperative is to understand what the desired outcomes are, and then find out which cloud providers will support them in ways that have all the enterprise needs covered: availability, predictability, manageability, flexibility and reportability.
This last one is important. Only through having constant reports on how well the platform is performing can the future be predicted – and adaptations to the platform be made to ensure that key performance indicators (KPIs) are still met.
Believe me – as long as all these areas are met, the resulting platform will almost certainly be cheaper in the long run; even if the up-front cost is higher than what other platforms are offering.
Negotiating cloud contracts
There is one other aspect of falling costs that any prospective cloud user must bear in mind. This is that the price you agree now is not going to be the price in a few months’ time, and certainly won’t be the same by the end of a long contract.
Many customers are watching as the three-year costs they agreed are being undercut for new customers as prices fall, but they are contractually bound to keep paying the higher price.
This effectively enables the cloud provider to use these “prisoner” customers to cross-fund new customers, allowing the provider to undercut its competitors – good news for the prospect, but bad news for the existing customer, who could find themselves cross-funding their own competitors in the market.
The price for cloud services will bottom out and may start to rise
When negotiating a cloud contract, by all means aim for cost predictability, but build in review points as to how much core resource capabilities are costing you. Even better, write a contract that does not use defined costs for the resources, but works on a percentage of list price – maybe that as you are signing up for a three-year contract, the cost of CPU, storage and network is 50% of list price to you throughout the contract term. This allows your pricing to track the provider’s own pricing –so your organisation will still get the best deal it can.
As in the world of finance, you will have to accept that the costs can go up as well as down – at some stage, the price for cloud services will bottom out and may start to rise. However, analyst firm Quocirca estimates this is still some time away, and that organisations are best served by assuming prices will still decrease over the next 12-24 months, and are likely to remain relatively steady for a year or two after that.
Cloud computing is changing the way organisations view their IT platform, and the way their organisation operates. Chasing costs downwards could be the worst thing to do as un-optimised and poorly architected clouds could fail and bring the organisation down. Ensure desired outcomes are facilitated and that the enterprise’s IT requirements are met – and look at the overall lifetime value of the platform, rather than the upfront cost.