In the 80s and 90s, the most common software licensing model was the concurrent/floating license model. What's not to love?
- Install software on any # of devices
- License server will prevent more than <purchased quantity> from using at any time
The theoretical advantage is that enterprises don't have to buy licenses for every user, only for peak # of users (most at any one time).
Seems like the perfect solution…right?
- Always in compliance
- Easy to administer
That may have been the case in the 80s and 90s, but for the last few years, we've seen a large decline in this preference. From 2010 to 2012 in the enterprise version of the Software Licensing and Pricing Survey that Flexera Software holds annually with IDC, when asked what licensing models enterprises preferred, the preference for concurrent licensing declined significantly year over year.
- 2010: 60% preferred
- 2011: 40% preferred
- 2012: 24% preferred
I believe there are several drivers for this:
- Enterprises are prioritizing availability over costs (and getting shelf-ware in the process)
Determining "peak # of users" sounds easy, but:
- Enterprises need to budget a year in advance, so they have to determine peak use way ahead of knowing new hires and new projects
- If they are wrong, they might prevent some expensive resource (i.e. electrical engineer) from accomplishing a task and possibly missing deadlines etc.)
- Based on our review of many of our customer's log, they tend to over-buy, often quite significantly. Our guess is 25-30% on average.
- Some enterprises will want to reserve the licenses for particular projects or departments, because they are paying for those licenses This means that the story around "peak # of users" gets far more complex. It's "peak #" except that "Department #1" gets 10 of them and "Project #37" gets 30%. This can be managed locally –but it's often a nightmare. We've seen "options" files (the mechanism for controlling allocation of licenses) with 100s if not 1000s of entries.
- Publishers are not providing good licenses – making analysis almost impossible. Examples of "bad behavior" includes:
- Using same license version # from one product release to another
- Often enterprises need to keep older versions of software for specific projects. By not providing version #s on their licenses, enterprises have to purchase "peak #" across all versions, making analysis quite difficult.
- Using the same license feature across different products
- Enterprises then have to guess which users will use which products and then estimate peak use of the feature
So what's better? According to the enterprises that completed the survey from 2010 to 2012 is a shift towards:
- Device/named user based (+20%)
- "Larger" meters –site, financial metrics, enterprise, etc. (~10% - the question was not asked consistently year over year)
What licensing models do you think will gain favor in 2013 and beyond?