Research IT

Using Amazon Service Workbench for Remote Training

The pandemic has brought many challenges, not least conducting training.  Our Training team have recently used Amazon Service Workbench to teach Python remotely online.

We have continued to deliver a wide variety of training courses online during the pandemic, even though we have been working from home. The feedback from attendees has remained excellent thanks to a fortuitously timed switch in technology and a mutually beneficial collaboration between the Research Lifecycle Programme (RLP) Project S (Develop a service to manage restricted data) and the Research IT (RIT) training team.

Normally RIT courses would be taught on campus in a PC lab, but a new way of connecting students to these interactive collaborative courses was offered to course instructors at the start of 2020.  Using Amazon Web Services (AWS), centrally managed Cloud-based desktops were created. This allows course participants to access a standardised teaching machine via a web browser.

Course Instructor Viewpoint – Ann Gledson

Our RIT Python courses are interactive and require specific software installations and set-up, so that each exercise can be performed collaboratively. This was reasonably smooth when the course was taught face-to-face in an on-campus PC lab since the lab computers all have the same standard configuration. However, there were occasions where we also ran bespoke courses (e.g., to research groups) that were in other, untested UoM computer labs, or in just ordinary classrooms where each student had to bring their own laptop. This led to greater delays in setting up each different laptop, as we tried to navigate our way through different operating systems, versions and configurations. We sometimes ended up with several helpers stood around a student with an unusual laptop set-up, trying to problem solve why they couldn’t run a command. Then, to make things even more difficult, along came COVID and enforced working from home.

In early 2020, the RIT Infrastructure team offered to trial a new solution based on AWS whereby all the users connect to their own cloud-based desktop via a browser. This uniform desktop had all the required software and data files used in our lessons and, from the first trial session, everything ran smoothly. The time taken to connect the students with the tools required was halved, and any lingering connection problems could be referred to the infrastructure team to check their connection without getting in the way of teaching.  The students must digest a great deal of knowledge on these 1-day coding courses and it’s great that students are presented with a familiar interface and don't have to spend extra time and energy getting to grips with a complicated set-up. I also noticed that our participant feedback scores have remained high, possibly even improving.

Research Lifecycle Programme Viewpoint - Danielle Owen

So how did RLP Project S get involved in Python teaching? We have been working with AWS to test and develop tools that make Cloud services accessible to staff and students. One such tool, AWS Service Workbench, enables administrators to build workspaces with different software and compute specifications, which can be shared with end users to via a URL. Data can be uploaded to workspaces directly by users or administrators can upload data centrally and provide access to multiple users as with the Python course. When a user logs into the Service Workbench in their web browser, they can launch workspaces and view their data with just a few clicks. At the end of a session, workspaces can be stopped or terminated, keeping the cost to Research IT to a minimum. It was great to see Service Workbench in action during the Python course and the RLP team are pleased it went smoothly.

Course Helper Viewpoint - Anthony Evans

We used Amazon Service Workbench so that students could quickly create instances of the Python Jupyter Notebooks that we use for teaching. This had the advantage that each notebook could be reproduced identically for each student thereby reducing the need to support students who had different versions installed locally on their laptops. The spreadsheet files were already made available on Service Workbench’s S3 storage so students did not have to spend time downloading the dataset needed for their exercises. It also had the advantage that students did not need to spend time understanding the underlying server setup and could get started with their work more quickly. Since it was a Cloud-based service, it could even be used smoothly on low spec machines, enabling those with slow laptops to participate in remote learning."

Participant Viewpoint 

So how did those who attended the training find the use of Amazon Service Workbench?

"The remote desktop and remote session over Zoom is working very well. I think this way of delivering training courses is quite effective even under the "normal" circumstance."  I-Hsuan Lin, PhD, Bioinformatics Core Facility

“The set up with the virtual desktop, Zoom, etc. worked really well. I was very impressed with how easy was to get the technology up and running, and how engaging it was to follow the exercises yourself in a virtual training course. Thank you." Alexandra Martin Hernandez, Market & Student Insight Officer, Market & Student Insight Team

All RIT training courses and resources can be found on the UoM online training catalogue.