But one group of vendors still seems to want to slow down the march of virtualization: the application vendors. OK -- not all the application vendor community, as the new batch of Software as a Service (SaaS) and cloud computing-based vendors are showing that virtualisation is not only possible but the best way forwards. The software vendors dragging their heels are represented by the big incumbents, whose monolithic applications are held back by licence agreements that are archaic in the virtual world.
What are the main issues?
It is still common to see licence agreements based on the number of CPUs or even cores that are in use. In a highly virtualised world, just how is a company meant to measure this? The emerging dynamics of virtuality means that an instance of an application may be using 10.2 cores at a given time, add an extra 1.6 cores as a peak appears and then drop down to 7.4 cores afterwards.
The software vendors dragging their heels are represented by the big incumbents, whose monolithic applications are held back by licence agreements that are archaic in the virtual world.
Clive Longbottom, service director, Quocirca,
It may be that an image is kept spinning ready for use either to meet a peak demand or for business continuity needs; this may be using a couple of physical cores, but it is not in actual use by the business at all for the vast majority of the time. It may also be that only part of the functionality of an application is used; other functionalities are unneeded or drawn from elsewhere (for example, from another application or from the cloud). Yet it is still expected that the organisation pays for all of the application's capabilities on an ongoing basis.
These software vendors are caught between a rock and a hard place. On the one hand, they have built up their businesses based on a model that the money men understand -- charging per nominal amount of computer power available, measured through counting the numbers of CPUs utilised. On the other hand, the users are pressing for changes to a more subscription-style pricing approach based on functional usage.
Unfortunately, these two approaches are at odds. Wall Street and the other bourses don't want to see enterprise applications broken down into functional components that people can not only choose whether to use them but also how to pay for them. A monolithic approach works fine in their minds; you pay a lump sum up front for the privilege of using an application that does a lot of "stuff" and then pay ongoing maintenance from there onwards. The lock-in to the vendor is high, and the financial model has worked for a long time.
Virtualisation, however, is forcing change. The cloud/SaaS software vendors have shown that a subscription-based model can, and does, work. They have also shown that a new approach based on providing core functionality, bolstered through the addition of external functionality using "mash-ups" and easy integrations of external functions with the core platform, provides a far more responsive technical platform to support business needs. The incumbents have to change, but it is proving painful.
In addition to dealing with the financial community, with their three-month event horizons based on quarterly financial reporting, the software vendors also have to deal with their existing salesforce, whether this is internal or external through partners. An average deal is generally based around the salesperson signing a deal and then getting a percentage of the capital cost as commission in their pay packets. Moving from licence plus maintenance to pure subscription suddenly changes that. A million dollar deal that may give the salesperson a $30,000 pay check suddenly becomes spread out over a period of time -- maybe as much as 3 years or more.
Even if the overall deal is the same -- say a $1m upfront licence costs plus 17% per annum maintenance leading to a total software deal over three years of $1.51m, or a 36 month subscription of $42k per month -- the problem is apparent. In the first case, the salesperson on 3% commission will get a lump sum of $30k, plus $425 per month. In the second case, the salesperson gets $1,260 per month -- not quite the lump sum of $30k they were hoping for. If they choose to leave one month after a big deal, they still get the commission; what happens if they choose to leave one month into a subscription deal? Moving a salesforce over from a commission-based approach to a subscription-based approach has proven difficult for the majority of software vendors and their channel.
So what can be done? If the deal is big enough, sit down and talk to the application vendor and get them to draft an agreement that suits both parties. Make sure that you have the capabilities to implement the software as you need within your virtual estate -- try and keep away from physical limitations completely.
If the deal is smaller, then look at the options. If there are open source or commercial options that work in a virtual environment, use these as leverage against your preferred vendor. If not, then see if a per-user approach is possible where the physical or virtual platform becomes immaterial. If this also is not an option, then there's little option but to use the Internet and social networking sites to try and drum up enough feeling amongst the existing user base for change. Do not underestimate how powerful this can be.
Virtualisation is ready for prime time, and the incumbent vendors and the money men have to wake up to the fact that this is a fundamental change to the market. While they struggle with the concept, choose those who understand virtualisation or mitigate any possible issues through intelligent negotiation.
Clive Longbottom is a service director at U.K. analyst Quocirca Ltd and a contributor to SearchVirtualDataCentre.co.uk.