Experimenting on distant labs and labs on the Cloud
Some reflections and details about our plans for distance labs in the face of lockdown and the use of Cloud platforms for such needs.
The accompanying video is at: https://vimeo.com/424393463
Experimenting on distant labs and labs on the Cloud Olivier Berger <[email protected]> 2020/05/30 Olivier Berger <[email protected]> Experimenting on distant labs and labs on the Cloud
CV Olivier Berger Research Engineer at Télécom SudParis, in Évry (France) Teaching Web apps development to our Engineering School students Senior Engineer, working on digital learning technology Member of the support team setup for switching to distance learning (CoViD-19) Free Software activist Interested in virtual labs for many years (FLIRT MOOCs, etc.) : https://huit.re/blog-labs Olivier Berger <[email protected]> Experimenting on distant labs and labs on the Cloud
A few issues Lockdown: no (or only partial) access to lab rooms Lab rooms = critical resources increasing promotion sizes 1 machine for 2 students (but without collaborative learning outcome) (partial) BYOD O hours : accessible H24 ? Heavy deployment Fabrication process of the lab setups Unsed machines (at night) External cloud oerings competition (Google Collab, Rosetta Hub, etc.): GDPR, cost, lock-in ? Olivier Berger <[email protected]> Experimenting on distant labs and labs on the Cloud
Some solutions Plan : Virtual machines Web browser Remote desktop access on physical machiens in our lab rooms, through a Web interface Lab platforms on the Cloud Olivier Berger <[email protected]> Experimenting on distant labs and labs on the Cloud
Virtual machines Been there, done that. . . Welcome to 2020 ! Well trusted solutions 2005-2015 Designing a virtual laboratory for a relational database MOOC, Olivier Berger, J. Paul Gibson, Claire Lecocq, Christian Bac, Int. Conference on Computer Supported Education) 2015 Lots of issues mainly performance of machines fabrication process We may do that in last recourse Olivier Berger <[email protected]> Experimenting on distant labs and labs on the Cloud
Symplifying hypothesis Zero install for students Universal client : Web browser UX problems ? I'll only mention Free and Open Source Software Olivier Berger <[email protected]> Experimenting on distant labs and labs on the Cloud
Aiming next september Remote desktop access to machines inside School walls Remote connection through a browser (authenticated), to graphical desktops Web Proxy, with : Guacamole VNC protocol for Linux lab room machines RDP protocol for Windows lab room machines or MeshCentral Preserved licenses (Windows, Matlab, etc.) Similar scheduling (Lab in room n on machine m, for 1 or 2 users) Mixed solution : alternating presence / distance Files saved in the user's HOME Transparent for ou IT dept (deployment, etc.) Minimal additional costs : machines already there and powered, open-source Olivier Berger <[email protected]> Experimenting on distant labs and labs on the Cloud
Aiming next september Example: MeshCentral Remote view really fast for distant tutoring (labs) Olivier Berger <[email protected]> Experimenting on distant labs and labs on the Cloud
Aiming next september Feasable soon ? Di culties : 2 x machine load ? before : 1 session for 2 students after : 2 sessions for 2 students (1 present, 1 distant) ? LDAP/CAS auth feasability Control of graphical sessions startup on machines via VNC/RDP UX, network bandwidth, etc. Security (MeshCentral code audit ?) ETA soon (september ?) Working on Guacamole in Évry MeshCentral deployment in Mines Albi Olivier Berger <[email protected]> Experimenting on distant labs and labs on the Cloud
Open Source lab platforms on the Cloud Longer term Cloud, cloud, cloud. . . Virtualizing lab platforms to standardize better Olivier Berger <[email protected]> Experimenting on distant labs and labs on the Cloud
Open Source lab platforms on the Cloud Standardizing infrastructure with Kubernetes Cloud Up to scale with the Cloud Kubernetes cluster (k8s) : exsting big machines in lab rooms additions server machines (private cloud) external cloud (possible in theory, GDPR, ...) ? Docker : industrialization reuse Olivier Berger <[email protected]> Experimenting on distant labs and labs on the Cloud
Open Source lab platforms on the Cloud Antidote (NRELabs) Networking labs : multiple virtual device environments, for every student Young OpenSource project Docker, k8s Integrates Jupyter TSP is already a contributor (O. Berger) https://nrelabs.io/ Olivier Berger <[email protected]> Experimenting on distant labs and labs on the Cloud
Open Source lab platforms on the Cloud Improvements/contributions wishlist Authentication Moodle LTI integration Saving work Evaluations, etc. Graphical desktops Olivier Berger <[email protected]> Experimenting on distant labs and labs on the Cloud
Open Source lab platforms on the Cloud Eclipse Che Development labs, Software Engineering k8s, Docker Eclipse foundation project Web IDE : Eclipse Theia Authentication (keycloak) Workspaces : saving work . . . https://www. eclipse.org/che/ Olivier Berger <[email protected]> Experimenting on distant labs and labs on the Cloud
Open Source lab platforms on the Cloud Labtainers https://nps.edu/web/c3o/labtainers Distributed as a VM Cyber-security labs Docker Evaluation / personalization mechanism (cheating prevention) TODO: Integrate it into Antidote Olivier Berger <[email protected]> Experimenting on distant labs and labs on the Cloud
Open Source lab platforms on the Cloud Project ideas TODO : Validate labs industrialization with Docker approach (quality, reuse, updates, documentation, etc.) Integrate Antidote and Eclipse Che : auth, workspace persistence, networking or development labs (or mixed) Integrate the Labtainers evaluation mechanism into Antidote Who wants to play with me ? Olivier Berger <[email protected]> Experimenting on distant labs and labs on the Cloud
Conclusion Following the industry's evolution : Agility DevOps (pedagogical) Virtualization / Cloud Olivier Berger <[email protected]> Experimenting on distant labs and labs on the Cloud
Expected bene ts Universal remote access: Web browser Scalability / exibility: Kubernetes physical platforms, container virtualization, VMs Removing constraints of physical machines Pooling operation costs 1 environment for every student (no longer constrained to 2 on each machine) Standardization (same experience for professors, students) Industrialization of the fabrication process (quality, reuse, pooling, . . . ) Agile deployments (pedagogical DevOps) Olivier Berger <[email protected]> Experimenting on distant labs and labs on the Cloud
Risks Technicity: Kubernetes, Docker (IT sta training, teaching assistants, professors) UX Dependence to networks quality Open Source projects dynamics Olivier Berger <[email protected]> Experimenting on distant labs and labs on the Cloud
How ? Achievable ? Eort Pooling Open Source Kudos: Laurent Guerby (Mines Albi) about MeshCentral Questions ? Contact me : mailto:[email protected] Olivier Berger <[email protected]> Experimenting on distant labs and labs on the Cloud
oberger
bernhardsvt
jamashita
hachama
cbtlibrary
signer
debug_mode
masakiokuda
mansuy
irocho
mineo_matsuya
tetsuyaooooo
bkeepers
jponch
ufuk
andyhume
cromwellryan
csswizardry
phodgson
maggiecrowley
iamctodd
cassininazir
tanoku
tammyeverts
