Software Metering – Navigating What it Means and How Much is Enough

By Russell Parker

Answering how much software metering is enough is complicated by the fact that “software metering” can mean a number of different things, particularly when looking at the differences between “desktop” vs. “server” software. Even when limiting what we look at to just “desktop software,” the variance is large. In some cases, “software metering” is used to refer to managing software licenses and their allocation. A more general use of the term refers to looking at software “usage” or “utilization” data found by a discovery tool, but even in this case, the specifics can vary wildly. At one end of the spectrum, “usage” can be defined simply as the last time a software application was executed. At the other end is a much more complex and granular approach which includes detailed information about how long an application was open and/or active, as well as potentially looking at various resources consumed.

How Much Data to Collect

Discovery tools can collect a wealth of data and, while much of it is potentially interesting, not all of it is actually useful or needed. It is easy to get caught up in a complex analysis when a simple one might provide the answers we are looking for. So, when it comes to software metering, how much data is enough and when do we have too much? The key to finding the answer is thinking carefully about the question we are going to use the data to answer.

Identifying Unused Software

Companies spend a lot on software licenses, so naturally they are interested in whether it is being used or not. Identifying unused software deployments could help avoid unnecessary purchasing of software licenses or maintenance. Removing these unused installations before deploying the new ones would reduce the need for new licenses, but how long is long enough to make removal justified? Too short a time frame increases the likelihood of removing software which is going to be needed and also may run into issues with the licensing terms regarding removal and reuse. The key question of course is how to assess whether software is being “used” or not. A complex answer is usually more costly and time consuming to determine, so it makes sense to start simple. With desktop software just looking at the last execution date may get most of the way there for a fraction of the effort.

Example: A company has 1000 seats of Microsoft Visio Standard, 1000 deployments, and new requests for 300 more to be installed. Of the 1000 deployments 600 have been used in the past 30 days, 650 in the past 90 days, and 900 have been used at least once in the past year. The transferability period on MS Visio licenses is 90 days so the 90 day window is a relevant one. Targeting for removal those installs unused for >90 days looks good since this alone will free up 350 seats. Managing removal in a shorter period might be too cumbersome and the extra 50 seats may not be worth the additional effort since we have what we need for the moment.

The Viewer Advantage

If we are fortunate the simple time frame since usage question may get us all we need, but what if it does not? Can we use software metering data do better without having to dive in too much more deeply? It will depend based on the application in question, but one possibility is to look at whether there is a free “viewer” which allows documents to be read, but not modified. Since editing a document is much more time consuming, we might be able to use more detailed metering information to find cases where the full version is installed and a viewer might suffice. How well we can answer this question will depend on what data our discovery tool can give us, but we can at least get how long an application was open as well as how long it was active/in-focus. More detailed information, such as the total CPU time, RAM, mouse clicks or keystrokes, etc. may also be available, but the 80/20 rule again comes into play on how much detail is useful.

Example: As above, a company has 1000 deployments of Microsoft Visio Standard. However, the “last used” numbers are different. Of the deployments 800 have been used in the past 30 days, 850 in the past 90 days, and 900 have been used at least once in the past year. Just looking at the “last used” data only allows us to find 150 installations to remove, leaving us short 150 licenses. At first glance our usage is high, so more licenses are needed. But what type of usage? Looking at the open/active numbers for the past 30 days how we see: 200/100 > 48 hours, 300/200 > 16 hours, 600/400 > 4 hours, 400/600 < 4 hours, and 150/150 with 0 usage. Those who had it open and active for less than 4 hours in the past month were likely simply looking at documents and not editing them. This gives us 600 candidates for replacement with Visio Viewer. Since we only have 300 new requests at the moment we might be less aggressive and initially go for the 400 who had Visio open at all for < 4 hours. Addressing the Visio Viewer question to the 300 new requests to make sure they need to edit Visio documents not just read them would likely also be worthwhile.

Navigating the Sea of Data

Discovery tools bring back an overwhelming sea of data, but we must pick and choose what is relevant in order to extract the wealth of actually useful information buried within. Software metering data is just one small subset of what gets brought back, but it is one where people often seem to jump to “more is better”. There are hundreds of ways to potentially make use of it, but sometimes keeping it simple can get you what you need. Even when it does not, answering the simple question first will most often allow you to better frame the more complicated one.

About the Author

Russ Parker is the President of Golden Ratio, Inc.