The following is a brainstorm from Greg Hudson about Project Minerva as refocused around a web-hosting type
service for small-scale applications.================================== Angle #1: MIT infrastructure
What interesting and Institute-specific infrastructure do we have?
- Identity management (authentication, namespace, MIT IDs, WebSSO,
etc.) - File storage and sharing (AFS currently)
- Institute data storage (Data Warehouse, Moira, LDAP)
- Messaging and presence (email, XMPP)
- Printing (maybe relevant, maybe not)
What can set apart the Minerva hosting environment from random
commercial offerings is ease of access to MIT infrastructure.
Angle #2: The typical small-scale application
Borrowing from Wednesday's conversation, a department wants a small
application for reviewing TA applications. The basic requirements
would be:
- Accept applications via a web form
- Don't lose them
- Attach comments to them
This application is interesting because it requires about 1% of a
server and can probably be implemented in a couple hundred lines of
scripting, making it a great candidate for virtual hosting and a
horrible candidate for any of the services IS&T offers (unless DCAD
offers something I don't know about). It's also interesting because,
with a good MIT-specific toolset, it could be extended to:
- Authenticate applicants against the MIT namespace
- Cross-reference applications with institute data (about the
applicant and about classes taught by the department) - Indicate whether an applicant is online
- Send IMs to department staff when applications are received or
when new comments are created - Send IM or email feedback to applicants after processing
- Possibly even create printed reports about applications
all relatively easily within a scripting environment and without
ascending to the scale of a Project.
Angle #3: Athena
What's cool about Athena is:
- It's really easy to share stuff and to keep data safe
- It provides a rich scripting toolset to reach MIT infrastructure
- It's user-extensible
- It's universal (sort of)
- To some people, it feels like home
What's not cool is:
- It's based on marginal technology
- It's a desktop environment, which most people already have
- As a result, it has only a subcommunity interest
- Whatever you create on it is only useful to that subcommunity
A web hosting environment can be created using mainstream but still
open source technology (actually, there are too many choices there,
which is a pitfall). We can make it cool in most of the ways Athena
is cool. We can provide the tools to reach MIT infrastructure. We
can combine it with a project-hosting service to let users extend the
provided toolset with their own creations. It's obviously universal
since it's not tied to a particular client machine. And while it can
never be home in the same way a desktop environment can be, it can let
suitable motivated people create their own homes-on-the-web, which is
something people do all the time (livejournal, myspace, vanity
domains, etc.).