Main Nav

Message from

All - I am working on developing justification for departments requesting computer lab growth and I wanted to 1) confirm the information I am processing is correctly focused and 2) compare to other institutions to determine at what threshold percentage of utilization justifies growth. I am also curious to see how other institutions determine when to increase the size of a computer lab; e.g. based on enrollment projections, class scheduling, etc.? Here is what I am doing: From our Labstats system, I am exporting hourly (from the hours of 8am to 8pm) utilization (logged in and not turned on) from 1.1.2012 to today for Monday through Friday. Where there is a record of no utilization no row is returned; i.e. I don't have any data points that contain 0% utilization, but some that contain .05 (1 computer out of X in a lab). Typically I have at least 4000 data points when I generate my results. From the above information, I am then bringing into SPSS and generating boxplots and descriptive statistics. Typically I see an average of 15% to 20% utilization per computer lab, with a standard deviation of 15% to 20% utilization. I am also seeing a lot of outliers, as expected, which are instances in which the rooms are being used for instruction. Attached is an example result set for five computer labs that belong to the same department. ********** Participation and subscription information for this EDUCAUSE Constituent Group discussion list can be found at
Screen_shot_2012-04-04_at_10.10.22_AM.png41.37 KB
ATT00001..txt472 bytes


Jason, In the way your doing it, we look at the times/instances when we're at 100% utilization = 0% availability if there is someone in the queue. Of course we don't know, even if every computer is in use, if there is someone waiting to use the machine. We assume that if there is 100% usage (0% availability) - that people "could" be waiting. To state what may seem obvious, we only count the functioning machines...if a machine is down, it's not counted in the percentage. We export every couple of minutes (vs. hourly) and also use Labstats. If you find that certain labs are at 100% usage during certain times, days, etc. you can decide if more machines would make a difference. There are of course many considerations: ergonomics, size, location, access hours, etc. I don't think you'll find a standard for "what threshold percentage of utilization justifies growth" - but if you can determine how often you are at 100% utilization, you can then use it as input to the discussion and to the decisions you'll need to make. Marty
I like Marty's approach. I'm curious as to why you're limiting your analysis to M-F 8am to 9pm. Is that when the lab is open? At CMU, typical high volume times were late at night or on a Sunday night. Also, keep an eye on trends throughout a semester and think hard about what these trends mean to IT. We had one lab where students notoriously camped out during finals week because they were doing complex rendering for end-of-semester projects. The solution here was not to buy more computers, but to investigate a rendering farm.