Cloud Computing Myth 7: Cloud Only Makes Sense for Large Deployments

In recent weeks there appear to have been an increasing number of articles written about cloud computing myths.

158967_big_van_little_van_paris_stock.xchgIn recent weeks there appear to have been an increasing number of articles written about cloud computing myths. I am not complaining about that. In fact, I have written a few myself. And I am hopeful that this is another sign of maturing technology and a continuing increase in people’s understanding of cloud computing’s benefits, and caveats.

Mary Branscombe’s recent post “The six myths of cloud” made me think of another popular myth: “Cloud computing only makes sense for large deployments” (or when there are large surge requirements).  This myth sort of begins where her first myth ends, and it raises some important issues.

One of the key advantages of cloud computing is that it can lower the cost and risk of experimentation with new business models, innovative products, or research and development, while also reducing the time and effort required to create the environment for the experiment. This key advantage can exist at any scale. In fact, this is one reason that cloud computing can enable smaller businesses to compete with larger enterprises and, in some cases create disruptions that make life very difficult for their larger competitors. Once an opportunity has proven its value, cloud computing’s elasticity can enable them to scale to take advantage of their disruption. (Incidentally, those same opportunities exist for even the largest of enterprises, though many have been slow to take advantage of them.)

When comparing the cost advantages of an on-premise solution versus a cloud solution we need to think beyond the cost of the servers we would have purchased. There are many other on-premise costs, for example:

  • What is the cost of maintenance and support for the hardware?

  • What is the cost of software license and support?

  • What kinds of skills are required? Systems administration? Backup and recovery? Security?

  • What skill level is required? The higher the skill level, the higher the cost.

  • Can current employees fulfill those roles? At what cost and opportunity cost? Will they require training? At what cost?

  • Will the employees ever eat, take breaks, or go on vacation? Then you will need at least two people capable of fulfilling each role.

  • Will anything malfunction? Will users have questions? Likely. Do you currently have a service desk infrastructure? Do you have robust IT service management processes? You’ll need those too.

  • If you require backup, must you purchase hardware (or a cloud service) for that?

  • Is the network infrastructure adequate? Will it require upgrades? Will new equipment or wiring be required? At what cost?

  • What is the cost to manage and track all of this?

  • Do you have space in a facility suited for the equipment? Could you renovate? Would you have to build an addition or data center?

  • Is there sufficient power, cooling, lighting? If not, what is the cost to provide it?

  • How much time and effort will be required to make all of this happen?

Often these types of requirements are discovered after the servers are uncrated.

Any of these second order costs could, on its own, provide the cost justification for selecting a cloud service over an on-premise service; especially for a small or medium sized business. A good cloud provider should be able to create a robust, resilient, business-class computing environment at higher quality and much less expensively than a typical small business even at a scale of five or ten servers. And they would have it sooner, likely with a much better level of service and operational discipline than they could provide on their own.

As Branscombe suggests, and as I’ve recently written, things may not be as simple on the cloud side of the equation either. So you need to know what you’re getting there as well. (Protip: Do not assume backup and recovery is included, though it might be.)

And there is second order thinking required with cloud services as well. Cloud services are typically elastic, but you need to know how you are charged for your surges. (Are you charged based upon peak use? Average use? Per day? Per Month? Is there a premium per hour charge for complete flexibility? For guaranteed availability of new capacity?) There are, of course, also cases where larger enterprises can have the best of both worlds through private or hybrid cloud deployments. Though even they require second order thinking commensurate with the investment they will require.

I could certainly continue enumerating second order items on both sides of the discussion. Though my main point here is that these items exist at any scale; and so do the benefits. Even at a small scale, cloud computing may deliver benefits to your business. It can reduce startup costs and time. It can reduce the risk of experimentation. It can segregate the new systems from existing systems (reducing risk to your existing business), and it can enable you to succeed (or fail) faster and at less cost. And, as Branscombe suggests, it can enable you to get started much more quickly.

In addition, it’s not only about cloud options being less expensive compared to your costs. It’s also about quality, service features, and skill levels. It’s about what provides the best business value, which is not necessarily the least expensive option. So, don’t count cloud computing out only because the scale is small. If you do not, you might be extremely happy if your experiment pays off.


“Big Van, Little Van, Paris” photo courtesy of stock.xchng

This blog is cross-posted at Pragmatic Cloud.

You can follow @GeorgeDWatt on Twitter.

Written by

George Watt

As VP of Corporate Strategy at CA Technologies, George has spearheaded initiatives enabling organizations to…

Published in

View this topic
  • James Holland

    This is great. Hooray for Disney’s imagineers!


    become a new brand in the share market research with its accurate research. Proven
    itself always right whether market is bull or bear. Last week all paid clients
    booked handsome profit in NIFTY, BANKINIFTY & STOCKS. Now for the coming
    week we expect more correction can come in NIFTY as the IRAQ issue is getting
    more tense, If it happens more then you will see a sharp fall in all world marketNSE BSE, STOCK TIPSbecause as we know all world run on
    crude & most of the crude comes from IRAQ. So be ready for a sharp fall so
    sell will be the best strategy for next week also. Traders can make a sell
    position in NIFTY around 7600-7650 with stoploss 7750 for the target of
    7300-7200.One can also make a sell call NIFTY 50 stocks as per NIFTY levels. You
    can also take our two days free trial to check our accuracy. For further updates
    you can visit our website.



  • king lear

    testing comment functionality, please do not publish this

  • Rachel Macik

    Love the personal pic :)

    • CAHighlight

      Thank you!

  • Plutora Inc

    This is a good case study. 2.3 sec’s off a login transaction is big.

  • Michele Hudnall

    While the analysts were hyping DevOps, I posted the oversight of not including security as part of that discussion as you are highlighting here. Instead of just talking DevOps, it should be DOS (what’s old is new again :-) – DevOpsSec. As a previous AppDev person, it’s the app, who’s using it, why and where rather than the device and having the service available.

    As you rightly point, out Security should be baked into the solution.

    Nice Post and Timely!


    • CAHighlight

      Thank you for your feedback Michele. Agreed – security cannot be overlooked. Appreciate your input!

  • Mitesh

    I would love a printed copy

  • Lars Johansson

    I love the idea of BYOID! This makes me choose if I am almost anonymous (with my Hotmail Nicname) or official with identity from an official organisation. My Identity Provider will attach identity with right level of LoA according to the need of the Service provider.

    • CAHighlight

      Thank you for your comment. BYOID has tangible benefits for end users and relying parties but it also has to be weighed in the balance with potential risks and liability concerns. It will be interesting to see how BYOID plays out in the enterprise.