When a business today opens an electricity bill it sees the amount of electricity used over the billing period and the cost per kilowatt hour. What the business doesn’t see is a separate flat rate allocated per appliance and to the various pieces of electronic equipment that have been deployed in the office. Imagine paying a flat rate for the electricity to power your air conditioning, but only using it heavily for 2 months a year; commercially it doesn’t make sense. Yet we are willing to accept the flat rate model and the complexities that go with it when we buy and deploy cloud IaaS.
[easy-tweet tweet=”Should we be paying for #IT and #cloud the same way we pay for utilities? ” user=”comparethecloud” hashtags=”utilitycomputing”]
Is the current cloud infrastructure (IaaS) market, delivering an easy to use, consumer friendly service, similar to common utilities? Or should we be looking at the benefits of an alternative commercial model, utility computing, which differs from the widely used cloud IaaS model.
When comparing IT infrastructure to a utility model of delivering a service it is important to define what we mean. Standard utilities such as electricity and gas are delivered in a model where we, the consumer, only pay for the actual amount we use. Generally, this is considered to be the fairest way to buy utilities, as well as other goods and services such as food and commodities. All associated costs for the service, such as production and delivery, are included in the amount we pay at the end of the billing cycle. The key to the success of the commercial model is there is a single unit of measurement that enables the service provider to measure, and charge for the amount of resource that we have consumed.
the single unit of measurement is what makes the utility sales model successful
This model essentially removes the consumer from any of the responsibilities and financial investment into the supply, production and maintenance of the utility. The consumer is only concerned about the unit cost, the amount of resource used, and the reliability of the supplier. With regards to changing supplier, all suppliers use the same unit of measurement, and are therefore easily comparable. The complexities of managing supply and demand are removed from the consumer and the buying process is simplified to make the resource easy to use.
Let’s look at how this model translates into the world of IT infrastructure and the cloud IaaS market.
Traditionally organisations/ the consumer owned and invested in the technology and people to deliver their IT. The organisation took on the responsibility of managing supply and demand, so they could deliver data and applications to the business. They also, inadvertently, took on headaches such as keeping pace with changing technologies, recruiting and skilling specialist staff, scaling to meet demand, financing ad-hoc capital investment and building specialist facilities, the list goes on.
[easy-tweet tweet=”Should there be a single unit of measurement for #Cloud services? ” user=”comparethecloud”]
This is not a simple model for the consumer to understand, and therefore control. As a consequence, costs and delivery standards were, and still are, wildly different between organisations.
The cloud era has brought about IaaS and more of a utility aspect to infrastructure delivery. Cloud Service Providers (CSPs) are specialised and the consumer pays on an opex basis, no ownership of assets and no responsibility for supply and demand. However, it is not common practise to pay for IaaS as a true utility, or in other words only pay for the resources that have been used.
if you deploy a 100 IaaS instances and most of them are over provisioned for the workloads they support, then you are wasting a considerable amount of money
Most CSPs sell their cloud IaaS instances at a flat rate depending on the amount of resources (CPU, storage etc) allocated. It doesn’t matter how well utilised the instance is over the billing period the customer pays the full amount. Therefore, if you deploy a 100 IaaS instances and most of them are over provisioned for the workloads they support, then you are wasting a considerable amount of money. For context, we see far more over-provisioned servers, than we do under-provisioned.
The other problem is variation in sizing (resources allocated) and flat rate pricing models for the IaaS cloud instances. Because each CSP’s rates and instance sizes differ, it is an overly time consuming and complex task to work out who is offering value for money, and then to forecast your future spend.
The cloud model has changed infrastructure delivery for the better, however it is still not delivered in a consumer or business friendly model. What both the traditional investment ownership and cloud IaaS models have in common is that billing units are fixed logical containers of resource. The logical containers have just progressed from physical server, to virtual server, to cloud instance. If we want to be more efficient with infrastructure usage and IT spend, the industry needs to look to our common utilities for inspiration.
to be more efficient with infrastructure usage and IT spend, the industry needs to look to our common utilities for inspiration
For a utility computing model to be effective in the IaaS space there must be a common unit of infrastructure measurement. 6fusion’s Workload Allocation Cube (WAC) is a patented algorithm used to formulate a single unit of IT measurement, that combines the 6 compute resource metric readings together. A thousand WACs is equal to one kWAC hr, similar to gas and electricity utilities.
This unit of IT measurement is already in use to facilitate trade on a utility computing spot exchange called the Universal Compute Xchange (UCX). The exchange has a number of vetted CSPs selling their IaaS products through their online trading platform. The best way for customers to buy via the exchange is to engage a registered broker.
The broker role is a relatively new one in IT but not uncommon in other industries. The broker will be a member of the exchange and has access to better pricing than non-members. Their role is to facilitate trade between the buyers and seller. However, as this is IT, a buyer may have specific needs for their workloads, so the broker will need to understand the requirements and purchase units with all the necessary assurances in place. The requirements organisations currently have like performance, security, compliance and service levels will still need to be addressed.
[easy-tweet tweet=”What the #utilitycomputing marketplace gives the customer is pricing transparency and usage based billing”]
What the utility computing marketplace gives the customer is pricing transparency and usage based billing, so finance and business stakeholders can see where their IT spend is going. The simple commercial model also enables benchmarking and forecasting of future consumption and spend. This model seems superior to the currently more popular flat rate per IaaS instance, where the price is calculated on the resources allocated. Benchmarking and forecasting future deployments and spend is inherently more difficult in the flat rate model.
If the future of IT infrastructure is to be treated as a utility and a commodity that is on tap to drive innovation and customer engagement, then consumers must demand a more commercially savvy charging model. It’s time to consider utility computing and the commercial benefits it can bring.
It’s time to consider utility computing and the commercial benefits it can bring.