The Rabbit Hole

Considerations and reflection on technology and IT strategy

linked-In Follow on TwitterRSS Feed

Cloud Computing value for consumers

As we have seen in the previous article, consumers are supposed to gain multiple advantages from Cloud Computing (flexibility, cost reduction,…) but let’s see in detail what are the opportunities of Cloud.

Small and Medium Businesses (SMB)

SMBs are probably in a position to get the greatest advantages of the Cloud paradigm. Indeed, as of today, SMBs usually can’t afford Enterprise class solutions. Such high end solutions often require investments that SMBs can’t bear or subscriptions on volumes they can’t reach. Furthermore, in the case of starting businesses, it is also obvious that investing on an infrastructure, or subscribing services on several years is a real burden that increase the financial pressure.

For instance, ERP are quiet expensive to implement, so expensive that the economic equation is not always positive for SMBs. But with a SaaS model, an ERP becomes affordable. With a real pay per use model and no infrastructure expense, cloud can open such solutions to businesses that weren’t able to use them before. Because Cloud is more flexible than traditional outsourcing and with a finer granularity, it will be a very interesting field for SMBs.

In a general manner

Enterprises usually already have an IT department with IT facilities. They potentially have access to every type of solutions. So, what value Cloud can bring for this type of actor?

The market is of course underlying that Cloud improves IT resources utilization and, as a consequence, lowers the utilization prices. This is of course an advantage, but Cloud also brings better capacity management, as Amazon underlined in its now famous Capacity versus Demand Curve:

Capacity vs demand

Classic capacity is expanded through acquisition expenditures. In the most favorable case, resources can be provided through outsourcing but this type of service requires sometime mid to long term engagement. As you can see on the curve, Classic Capacity never really matches the demand. It is either above, leading to overcapacity, or under. In this situation, either you don’t get the full benefit from your IT investments or you may have direct impact on your business.

Each time an under capacity situation is encountered, if the upcoming demand can not be precisely predicted, the decision to add capacity is very risky.

Because Cloud brings flexibility and a virtually unlimited capacity to consumers, demand can be matched precisely.

But focusing only on an infrastructure optimization will only result in costs rationalization and improved efficiency. This would just be missing the global picture ; Capacity and Utilization are only IT performance indicators. They are not related to Business Value and Return On Investment. To define Cloud’s value we should examine how it could be beneficial from a business perspective.

Shorten service provisioning cycle

Who has never experienced project delays due to the  late provisioning of  the hardware, or  network? In a traditional IT environment, one generally follows a design – procure – setup cycle. This cycle usually last several weeks (months in the worst case). Furthermore, it is often  placed on the critical path of a project. Indeed, procurement can’t be launched until the design is over and soon after the design phase is over, you need to setup your environments for development and testing (as illustrated by the following).

Traditional provisioning cycle

In short, you nearly have no margin during this phase and any delay in procurement or setup results in a global delay for the project.

With a Cloud provisioning cycle, you dramatically reduce the time required to have an operational environment or service. What usually takes weeks or months can be reduced to days:

Cloud provisioning cycle

This reduces the risks on the planning and leads to a simplified development phase (as you no longer need to manage a long and complex early work). On the global picture, this enables to have a shorter project.

The benefits for the Business are obvious. The sooner a project is done, the sooner new functions are available. This enable to achieve greater agility and shorter time to market. This is particularly interesting for enterprises on a rapidly evolving market and who are looking for a competitive advantage.

Optimizing costs and simplifying ownership

As we already told at the beginning, Cloud enables to provision resources and licenses in a flexible way that avoids over capacity. But what could be the associated benefit?

Capacity is generally allocated to fit the peak demand, thus in other periods, resources are idle. The utilization ratio varies among studies. But you’ll read that it ranges from 5% to 20%. This is a huge source of optimization. Furthermore, a Cloud provider will buy large volumes of hardware and software licenses. This results in an improved service price for consumers.

Another important source of costs relates to administration and operations management. Depending on the service levels, operating a platform can be very complex and expensive. Sometimes, companies don’t reach the critical mass enabling them to have industrialized procedures and a high level of skills. On the opposite, a cloud provider has efficient management processes and a high level of skills. For a given level of service, providers operate at a lower price.

With Cloud, consumers will transfer outside the responsibility of matching capacity and demand. A paper from Berkeley University draws a very relevant parallel with the semi conductors industry to illustrate this interest:

Cloud Computing is likely to have the same impact on software that foundries have had on the hardware industry. At one time, leading hardware companies required a captive semiconductor fabrication facility, and companies had to be large enough to afford to build and operate it economically. However, processing equipment doubled in price every technology generation. A semiconductor fabrication line costs over $3B today, so only a handful of major “merchant” companies with very high chip volumes, such as Intel and Samsung, can still justify owning and operating their own fabrication lines. This motivated the rise of semiconductor foundries that build chips for others, such as Taiwan Semiconductor Manufacturing Company (TSMC). Foundries enable “fab-less” semiconductor chip companies whose value is in innovative chip design: A company such as nVidia can now be successful in the chip business without the capital, operational expenses, and risks associated with owning a state-of-the-art fabrication line. Conversely, companies with fabrication lines can time-multiplex their use among the products of many fab-less companies, to lower the risk of not having enough successful products to amortize operational costs. Similarly, the advantages of the economy of scale and statistical multiplexing may ultimately lead to a handful of Cloud Computing providers who can amortize the cost of their large datacenters over the products of many “datacenter-less” companies.

quoted from Above the Clouds: A Berkeley View of Cloud Computing

In short, Cloud will help consumers to concentrate on their core business, leaving production issues to specialized people and enabling to invest capital in other activities.

Moving from a CAPEX model to an OPEX model

Cloud is going to change the way of financing IT. As we have already seen, traditional IT mainly relies on CAPEX as you need to invest in capacity before using it. From a business perspective it means:

  • upfront investments: Capital must be invested to acquire hardware and licenses. This capital has a cost and cannot be invested elsewhere.
  • long ROI: return on investment can sometimes take up to one or two years. In a volatile economy, it might be impossible to forecast such ROIs in a reliable manner.
  • heavy risk: as in any investment, there’s a risk of failure. Up front capital investments may be partially or completely lost in case of failure.

Cloud enables to move from CAPEX to OPEX. No upfront investment is needed and OPEX is mainly driven by the demand. Instead of being tied up in fixed assets (IT resources) money is still available to be invested elsewhere (or less money is needed to launch a service). If you consider that capital has a cost (Weighted Average Cost of Capital), Cloud helps to improve revenue or to use capital in the more efficient way. And if we compare with traditional IT you have:

  • zero or minimum upfront investment: capital is better used.
  • quicker ROI: as you don’t need to amortize fixed assets ROI is improved.
  • minimized risks: as you need less capital to launch your service you minimize the risk in case of a commercial failure.

Challenges for consumers

It is hard to be exhaustive when talking about Cloud challenges, but here are some of the most important.

Security and regulations compliance

Although Security was seen a major show stopper, people now tend to consider that many provider are able to guarantee higher security levels than most in-house IT systems. But Security is still a major concern for enterprises, especially regarding sensitive data hosting.

There a lot to bet that Security is going to be managed as a service level in the years to come. This will surely simplify things. But regulations will probably still be an issue for some time. Indeed, Data Privacy and Protections laws require in some case that private data stay on a national territory and in some cases these data can’t be hosted outside the enterprise. Providers begin to sell specialized services for such requirements, like localized datacenters with special access policies, but as of today, this still raises obstacles to cloud adoption.

Dealing with the illusion of reduced complexity

One of the great promises of Cloud Computing is a reduced complexity from the point of view of the consumer. As far as we talk about Software as a Service like Mail I can’t disagree: every end user can easily subscribe a new mail account without any help from an IT department. But as soon as you deal with Platform or Infrastructure as a Service, it is hard to imagine a business user requesting the provisioning of a new system image and configuring its fresh Oracle database installation for the new application development project.

This is what I would call the illusion of reduced complexity. For sure, Cloud makes things easier but not to the point where the IT departments are just going to manage their contracts with cloud providers. Cloud is going to bring new challenges that will have to be addressed:

  • multi-provider environment
  • seamless integration with internal Information System
The never returning resource

Cloud relies on a simple paradigm: resources or services can quickly be instantiated and released. Provisioning a service that will never be released breaks this paradigm to some extend. Two points must be underlined here:

  • Owning a resource to process a constant load on a long term is more cost effective as paying for the equivalent cloud service. Indeed, Providers need to make money to have a viable economic model. Consumers must use Cloud computing for appropriate loads.
  • Keeping an unused Cloud service is the worst. Consumers must enforce strong policies to ensure that unused services are returned to the cloud. Otherwise costs are going to explode.

This means that IT Governance has to evolve to take Cloud into account (choosing appropriate loads, subscription usage tracking, …) otherwise cloud adoption could simply fail.

Facing the lack of standard and interoperability

As of today Cloud Computing lacks of standard. Each actor is proposing it own formats and interfaces. Consumers face the risk of a lock in effect. Once data are hosted outside, a consumer may become dependent of a proprietary system whose cost may escalate or terms of service may change unilaterally.

If no standard emerge sooner or later, consumers will have to maintain an abstraction level between them and their provider to manage interoperability.

How to start?

As we have seen, there are many opportunities and challenges. If you’re interested in moving to Cloud as a consumer, you should first assess what are your current IT capabilities and what could be the potential impact of a cloud adoption. In a second time, you should assess the impact on your business model. This second point is really important. If you see Cloud as a simple opportunity to reduce your IT costs you’re missing a major part of this disruptive change. Organizations that are going to define a Cloud strategy to improve their business model or to create new products and services are going to create new markets or take a decisive advantage against competition.

Starting with a pilot is a good choice. Choose an appropriate workload type and a business imperative where Cloud benefits can be compared to a traditional IT. Make your choice among the different types of Cloud (IaaS, SaaS, …) and analyze your pilot experimentation to confirm your global strategy.

Laurent Chades


Posted on January 3rd, 2010 by Laurent Chades

Ok ok, I know that IT and SOA governance have been quite fashionable words. But let’s stay away from hype in this short note.

Sometimes, integrators don’t see the need for governance. Recently I even heard someone saying ‘SOA governance are boyscouts’ practices ! We don’t need this‘. In his defense,  I must admit that we have heard and red so many things (too many?),  that governance has been completely discredited. Furthermore, integrators are often present for a limited period. They deliver the system their customer ordered and their action is over. They not always see what a system without governance can become.

From a recent experience, I would say that SOA governance must be implemented directly by  customers (or with some help from third party consultants).  First, this is a way to ensure that the implementation and management of the system fits in the initial vision. The second point is that SOA governance is often a question of  arbitration between actors of the enterprise. Considering this, an integrator, as an external actor, may not have sufficient political weight to be efficient in this role due to the following points (not exhaustive ; feel free to add yours in the comments):

  • An integrator may not fully have access to the sponsor whereas others actors from the enterprise may have.
  • As I’ve seen sometimes, an integrator may have direct or indirect stakes in another project involved in a decision making that could prevent him to follow the governance principles

Another common pitfall I’ve met is the illusion of SOA governance. In this case, the project had a good design method, that ensured traceability of the functional requirements from logical architecture to the code delivery. This is a good practice, and the technical leader argued that since it was possible to track the impact of a change in the code, it was enough to avoid the definition of services governance. But SOA and services governance are different things, good design and development practices don’t substitute to governance (and vice versa). SOA governance is not about managing the implementation and the internal design of services. It focuses on the services availability to consumers, evolution decisions and the impact analysis on actual consumers, decision to expose or withdraw a service, management of the service levels and capacity planning …

No matter in which case you are, if your governance process is not efficient or doesn’t exist, one day or another, someone is going to say ‘Oh my god!! We had a really clear vision of what we wanted and how it was supposed to work, how did we get here?‘.  But ‘efficient’ doesn’t mean necessarily heavy. A good governance process must be appropriate. By appropriate, I mean 2 things:

  • Setting up governance must be an iterative process. With you first integration project introducing the need for SOA governance, you should only concentrate on what’s relevant at this first level. The principles you define must focus on your immediate problems: service life cycle definition, service versioning policy, … Keep complex things  for a later iteration, for instance, you won’t meet governance issues concerning 2 different service consumers asking for 2 different evolutions of the same service if this service does not exist yet. Concentrate on the first delivery of the services to their consumers. This is more appropriate and will enable you to reach a good maturity level of your first principles ; they will be the foundations for the next iterations.
  • Don’t get lost in over-sized tooling ; Governance is not ‘tools’. Deploying and customizing complex tools if you’re in the early stage of your SOA transformation could lead to extra costs. Indeed, your governance process will probably change a lot at the beginning and you maybe won’t have a lot of services to manage at this stage. Wait to have a stabilized version of your governance, this will prevent you to implement something you are going to remove. Furthermore, if you are not familiar with SOA principles, you’ll have a better idea of the help a tool should provide once you have made your first iteration and your process is running smoothly.

As a conclusion I would just say don’t neglect your governance process. It’s not required to be huge and overwhelming, keep it simple and rational.

Laurent Chades' Blog

  • Tags

  • RSS News

Copyright © 2010 Laurent Chades. Theme by THAT Agency powered by WordPress.