Cloud Maintenance

McKinsey tells us that the “Cloud” is over hyped and users can get better TCO through “aggressive virtualization”. Google responded , saying that there is more to the Cloud than just virtualization. They point out that very large scale hardware infrastructure is hard to do, the “Cloud” is more about software than hardware, and finally, that the Cloud delivers a constant stream of innovation—something that is very hard to do by just virtualizing your applications.

McKinsey tells us that the “Cloud” is over hyped and users can get better TCO through “aggressive virtualization”. Google responded, saying that there is more to the Cloud than just virtualization. They point out that very large scale hardware infrastructure is hard to do, the “Cloud” is more about software than hardware, and finally, that the Cloud delivers a constant stream of innovation—something that is very hard to do by just virtualizing your applications.

Clearly, we find ourselves in violent agreement with Google. The McKinsey report is full of the type of “facts” that the witches in Macbeth would be proud of! To pick one statement almost at random:

It is hard to argue with, except that the authors don′t bother to point out the obvious logical conclusion: Clouds will be used by all large public and private enterprises because all disruptive technologies in the history of computing start this way. If I′m not being clear enough, try resetting the date to 1992 and replacing the word “Clouds” with the word “Personal Computer” and pretend you are sitting in a conference sponsored by Digital Equipment Corp. But don′t take my word, ask Flextronics or any other of our 80 enterprise customers why they chose the Workday cloud.

It′s the Maintenance, Dummy!

Building on the Google argument, we′d like to add a new dimension to the debate on the advantages of “Clouds” — The Enterprise Maintenance Vortex.

We bet that if you got a 100 CIOs in a room and asked them to rank the “riskiness” of several major IT initiatives, that you might come out with a chart that looks a little like this:

  • Choosing, implementing and maintaining hardware is the bread and butter of most IT organisations – they are really good at it, and it is a core competence. I don′t think that any CIO would class this as a risky activity.
  • Creating, maintaining and upgrading some custom written software is a “moderately risky” activity. Most CIOs would like to avoid it, but it is a activity that is carried out by all IT shops and is something that they have learned to manage well over time.
  • ERP selection and implementation is probably “high risk”. Whilst hard data is difficult to come by, studies have suggested that “70% of ERP implementations have failed, have come over budget or have been seriously postponed”. I don′t think there is a CIO in the world who would look forward to an ERP implementation.
  • Finally, if ERP implementation is “high risk” then an ERP Upgrade is “beyond risky”. An ERP Upgrade is in essence a re-implementation of the original ERP solution with the added risk that the system is already up and running. It is open heart surgery on databases with several thousand tables, untold lines of custom code, and a whole load of integration code! ERP vendors are so sensitive to the risk associated with these “Upgrades” that they have dedicated “Upgrade Labs” to help customers with this problem.

Maintenance Economics

The core problem with the “aggressive virtualization” as proposed by McKinsey is that it keeps customers, and their IT departments, in the enterprise software maintenance business. Not only are they in the business of maintaining and upgrading software they′ve written themselves, but the need need to maintain and upgrade someone else′s software. (I′m reminded of the “Someone Else′s Problem” invisibility joke from the Hitchhikers Guide to the Galaxy—Someone Else′s Problems are so big you can only see them from the corner of your eye!)

However, the situation is a little worse than this. Because all the risk is on the customer to do the upgrade, the ERP vendor has no incentive to make this problem easier for the customers — rather they have the main incentive to add as much new functionality into the product as possible, making it more attractive in the market and, perversely, even harder for the customer to upgrade. (All the new functionality means even more changes to all those database tables).

Contrast this with the “Cloud” economics: All of the cost and risk of the upgrade is the responsibility of the vendor, therefore the vendor has a massive incentive to make the cost of upgrading as small as possible. At the same time, the “Cloud” vendors are competing against fully featured ERP solutions and have the market incentive to bring out as much new functionality as possible. This is the core reason why “Cloud” vendors have created software architectures that handle change—something that ERP vendors have never really figured out.

All of this ends up in one place. If you want to talk about the economics of “Clouds”, be they “Private Clouds aka Aggressive Virtualization” or real “Public Clouds”, you have to compare apples to apples. There really is no point, as McKinsey have ham fistedly done, in just looking at the hardware costs You need to put all the costs of ownership on the table: hardware, software, maintenance, upgrades and innovation. When it′s all on the table, then real “Cloud Computing” wins out every time.

More Reading