gwsystems / sledge-serverless-framework

SLEdge: a serverless runtime designed for the Edge.
GNU General Public License v2.0
107 stars 16 forks source link

University of Sydney Honors Project #356

Open richard875 opened 2 years ago

richard875 commented 2 years ago

George Washington University Systems Lab:

Hey guys, my name is Richard and I'm an honors student at the University of Sydney (Bachelor of Science with Honors majoring in Computer Science). My honors project is about WASM and particularly WASM running on IoT. I reached out to Sean (@bushidocodes) earlier and was advised to open an issue here. This is a one-year project and my degree finishes in November 2022.

Since this is a relatively new field of research, I'm having a hard time deciding on a research question as well as research directions. However, both my supervisor and I have agreed to commit to conducting my honors research based on the Sledge and aWsm papers.

It will be very helpful if I can be advised on some of the tasks and projects gwsystems team thinks are appropriate for a university honors year project and I look forward to working with the team. πŸ™‚

I have provided Sean with my research proposal.

P.S. Please feel free to visit my personal website: https://richard-lee.com/

emil916 commented 2 years ago

Hello Richard,

I am a second year PhD student in the Systems lab and currently mainly involved with the Sledge framework. Nice to meet you!

I read your proposal and it is pretty interesting. Good luck with that!

Can you please specify precisely what kind of assistance (or advising) you are going to need from our end? aWsm and Sledge installations should be pretty straightforward given the README instructions.

Emil

gparmer commented 2 years ago

You proposed to look at execution time for different applications in WASM, and perhaps understanding where overheads come from. I don't know what the requirements of the program are, but that is certainly an interesting avenue for investigation.

But I'm not sure if that requires Sledge. An interesting project might be to make Sledge support persistent computation, but this would take a lot of work from you to design it, and test it. Currently we only support serverless in which the sandbox is cleaned up after a request is formulated. A stepping stone to that might be to simply support HTTP 1.1 or higher with persistent connections. But both of these would require a lot from you to dive into the codebase and figure out what needs updating. We don't have the throughput to advise someone on this aside from answering specific issues with the current Sledge code.

richard875 commented 2 years ago

Thanks for getting back to me, the advice from @gparmer is certainly interesting. However, as mentioned that it would take quite a lot of work to develop and test it (I will finish my degree in a few months). I believe the assistance/advising I'm looking for is if I can be advised on a research project within sledge and aWsm that can 1: Help out the GWS team and 2: Able to fit in within the university degree's timeframe.

Cheers, Richard

gparmer commented 2 years ago

Depends on what your expectations for advising are. If you mean a weekly (at least) back and forth about details of the project, we cannot do that. We have no idea what your universities requirements are, and don't have much free time. If by advising, you mean us answering technical questions and problems you have here, we can certainly help with that, and are more than happy to. Generally, if you feel that you have a goal, and can dive into the system, we'll help you make progress when you run into issues.

Let us know how you want to proceed! We wish you the best either way.

richard875 commented 2 years ago

Thanks, all I'm looking for is a research topic suggestion. let's start with the suggestions in this thread and set a goal with them.

Thanks, Richard

richard875 commented 2 years ago

Hi team,

I think it would be a great opportunity to work on enabling persistent connections (HTTP 1.1) on the sledge framework.

Let’s get started on that! I can undertake most of the research by myself along with my supervisor at SydneyUni, however, in the meantime, would I be able to get a bit more context/details on the topic itself, for example the end goal, the requirements, and so on?

Regards,

emil916 commented 2 years ago

Hi! The questions you asked are pretty straightforward and have been answered in the Sledge and awsm papers, I believe. If there's any concrete question along the road, I will to try to answer.

Good luck diving in!

richard875 commented 2 years ago

Thanks @emil916, will update you guys along the way πŸ™‚