GianlucaMassimiani wrote > 1) Install an hypervisor (e.g. KVM) on each server > 2) Install Guacamole on the servers > 3) Connect to the cloud system through a web browser, being able to see > which servers are available > 4) Through the web browser, select a server and specify the software (for > example, the OS) and further specifications of the instance (virtual > machine) that I want to run on that server. > 5) Launch the instance on the server using the KVM, and use the instance > through the web browser > 6) When I am done, disconnect from the server (instance on the server > should > be destroyed)
AWS will not let you have hypervisor access on anything less than a $3500/month server. My org didn't want to pay for that overnight (since we are taxpayer funded and generally don't have users overnight), so we went with one or more M5's running guacamole as a container (AWS ECS using the official Guac containers at https://hub.docker.com/u/guacamole). We then use an out-of-guacamole system to create/start/stop/delete lots of T2/T3/M4/M5 instances. In reality, I think your entire use case is very similar to what we do at the Virginia Cyber Range -- spin up an instance on demand for a student, then stop or destroy it when it's done. We don't currently rely on anything in Guacamole to manage our servers/instances, and we use our own UI & back end to decide what to build, when to start/stop the VM, and prepare any customization required. Guacamole is _great_ at the RDP/SSH pass through once you get it working, and it's really helpful in school/government environments where installations might be restricted, or IT network admins are concerned -- though do consider that your process might create a hole which might bypass network filters. Good luck! -Lee -- Sent from: http://apache-guacamole-general-user-mailing-list.2363388.n4.nabble.com/
