Hybrid Learning Spaces effective use of your existing labs with Cloud Virtual Machines
So for the last 10 years the concept of Hybrid Learning Spaces has been growing across academic institutions this week I attend a Industrial Board where one of the key debates was growth of the department but how this was limited by computer lab faciilities and simply how many PC could be fitted into the lab,
Back in 1976, Maastricht University Library developed the concept of the Study Landscape to respond to students' needs. It consists of a combination of study spaces and learning resources, supervised by skilled librarians and with generous opening hours.
Many UK Universities have taken on this challenge two I know very well are University of Sheffield Information Commons and the University of Manchester Alan Gilbert Learning Commons. The Study Landscape contains the library and all the necessary facilities are concentrated in one building.
These building are an amazing blend of learning technology phycology which empower students to get stuff done, organisation like Google and Microsoft have resource spaces of similar format and characteristics.
So lets reflect this back on changes that have occurred in computing and the fact that computer labs have been computer labs for the past decade without any significant change other than the smart board.
So with the growing needs for flexibility and mobility, and students that are ‘digital natives’ shouldn't we consider how we redesign labs and resources to make these more effective what student needs.
So we got into the debate of do we really need to provide PCs in the Clusters or how can we make the spaces scale to allow us to have more students?
Well the key areas of the debate are
1. Software Provision – Licensed and appropriate to course requirements and needs so Windows, Linix, Specialised Software
2. The current life time of the PCs in the lab (i.e. Refresh cycle)
3. Capacity in terms of seats, table, power and connectivity (Number of Students)
4. Lab ownership is this a department or central IS own facility (Room Booking, Ownership)
5. CAPEx and OPEX costs of a lab rebuild or refresh and total cost of ownership (Budget Ownership)
So with all these factors many universities are moving to a hybrid learning approach with lab spaces and making them multidiscipline or multi use rooms.
A key driver to this is costs and expansion plans of the faculty or department.
During the meeting we discussed how the use of Cloud Services can really advance this, all students now have their own PCs, running Windows, Mac, Linux or a multiboot environment they connect to JaNET via EduRoam and Institutions have all made significant investments in WiFi.
Therefore the only blocker to actually getting students to utilise their own PCs is the cost of some products to license.
So what if the University simply took their existing desktop and virtualised the image and presented this as a virtual machine on the cloud.
Typical BSc courses are approx 280 credit points so if 1 module credit is equitant to 7.5 hours of teaching a student studying a UG BSc in Computer Science undertakes 2,100 hours or 87.5 days of learning/direct engagement. Off course they are expected to do far more in their own time.
So lets look at some simple cost models of using Azure Virtual Machines to support teaching labs using.
The way I want to present this is terms of OPEX costs for the faculty, department or IT Services. Microsoft has produced a great tools The Azure pricing calculator https://azure.microsoft.com/en-us/pricing/calculator/
We know from some scaling we have done across universities 25 users are easily supported by 8 Core 56Gb of Memory VM see the Data Science Virtual Machine post
We know Universities utilise Windows or Linux and some cases both so lets look at the costs for each
So if we look at the Azure pricing calculator at Windows Costs all costs as of 23/09/2016
We then take this cost and simply factor this up to the 2,100 hours so $6,83 x 2100 = $14,343 for providing VM to support 125 students based on 25 students per VM with a total of 5 VMs supporting 125 users.
So now if we look at Linux
We then take this cost and simply factor this up to the 2,100 hours so $4.69 x 2100 = $9,849 for providing VM to support 125 students based on 25 students per VM with a total of 5 VMs supporting 125 students.
No lets actually consider how many modules and time students actually need all that compute for so 2100 hours is approx 88 days of activities if we say a 1/3 of the time is based on lab exercises this equates to around 700 hours of compute time.
$4,781.00 USD = £3,672.67 GBP
$3,279.50 USD = £2,519.46 GBP
So under £4,000 of OPEX cost vs a CAPEX investment of 125 PCs at approx £530 per machine £66,520 plus associated maintenance, support, imagine prep costs. VM labs can be started and closed within 15 mins using custom scripts and Azure Resource Manager templates.
I know there are many factors but the purpose of the blog was show the offset of infrastructure costs from CAPEX to OPEX and the flexibility this provides in utilising your teaching and learning spaces to grow student numbers and student satisfaction.
Please share your thoughts and comments.