ago
October 26th, 2006, 10:23 AM
What about creating a Ubuntu-ZeroConf-HomeServer package as a spin-off of edubuntu?
"ZeroConf" in the sense that it uses ZeroConf and in the sense that it will work out of the box with reasonable settings.
"Home Sever" in the sense that its main (plugin-based) functionality should cover the needs of a typical home server. Including:
file server (indexed by tracker),
printer server,
multimedia streaming,
DVB/capture card streaming,
shared contacts,
shared calendar,
shared applications,
net-install,
thin-client server,
firewall/gateway/proxy server with A/V and antispamming filters,
centralized access restrictions to machines/web/applications,
user profiles,
backend for MythTV or equivalent,
import file facilities, that will also index and encode as appropriate (appropriate = using open standards)
one click backup to internal/external drive
....
All the above with a single, very easy to use web interface designed for inexperienced home users with only absolutely necessary options (for instance do not use groups, only users) all in plain english and well preconfigured. See Vista parental control for a good example of a nice interface.
Users should be able to install the HomeServer package and required plugins on one machine, and simply register or netboot other clients with this server in order to have a network ready in minutes.
The server should be modular so that it should be easy to add extra functionality. Each "plug-in" should be a deb package bundling required servers (as dependencies), good initial configuration, and web interface modules. The web interface should detect new plugins and show them in the menu. The focus of each module should always be on an extremely streamlined and intuitive user interface and good pre-configuration that should work out of the box. Advanced users can always access the daemons via webmin or directly via ssh.
This is useful today, also in homes that have clients from other OSs, or even for small businesses (a smallbusiness server package could be created as a twin project). It will become even more relevant in the near future when sub $100 machines like these (http://www.linutop.com/) will become ubiquitous and will be used as thin/fat clients as well as set-top boxes all over the house...
Reagarding set-top-boxes, the HomeServer could be a good occasion to standardize the backend. What I mean is that freevo, elisa, mythtv, etc could be split into a "standard-compliant" backend and a separate frontend, so that the same backend could work with different frontends. A MediaServer module might take such role, taking some of the load off the shoulders of such projects and avoiding code duplication. The shared folders are a natural place to store media files, metadata and querying could be handled via filesystem indexing (tracker), epg could be centralized, capture cards could be in the server as opposed to the client to get rid of cables in the living room, lots of possibilities.... This would make it easier to penetrate the set-top-box market as soon as possible, which is important since we can then spread open codecs and be in a better position to fight DRM.
Finally there should be a small client-side application available also for windows and mac to register and configure such clients with the main server and install required open codecs.
"ZeroConf" in the sense that it uses ZeroConf and in the sense that it will work out of the box with reasonable settings.
"Home Sever" in the sense that its main (plugin-based) functionality should cover the needs of a typical home server. Including:
file server (indexed by tracker),
printer server,
multimedia streaming,
DVB/capture card streaming,
shared contacts,
shared calendar,
shared applications,
net-install,
thin-client server,
firewall/gateway/proxy server with A/V and antispamming filters,
centralized access restrictions to machines/web/applications,
user profiles,
backend for MythTV or equivalent,
import file facilities, that will also index and encode as appropriate (appropriate = using open standards)
one click backup to internal/external drive
....
All the above with a single, very easy to use web interface designed for inexperienced home users with only absolutely necessary options (for instance do not use groups, only users) all in plain english and well preconfigured. See Vista parental control for a good example of a nice interface.
Users should be able to install the HomeServer package and required plugins on one machine, and simply register or netboot other clients with this server in order to have a network ready in minutes.
The server should be modular so that it should be easy to add extra functionality. Each "plug-in" should be a deb package bundling required servers (as dependencies), good initial configuration, and web interface modules. The web interface should detect new plugins and show them in the menu. The focus of each module should always be on an extremely streamlined and intuitive user interface and good pre-configuration that should work out of the box. Advanced users can always access the daemons via webmin or directly via ssh.
This is useful today, also in homes that have clients from other OSs, or even for small businesses (a smallbusiness server package could be created as a twin project). It will become even more relevant in the near future when sub $100 machines like these (http://www.linutop.com/) will become ubiquitous and will be used as thin/fat clients as well as set-top boxes all over the house...
Reagarding set-top-boxes, the HomeServer could be a good occasion to standardize the backend. What I mean is that freevo, elisa, mythtv, etc could be split into a "standard-compliant" backend and a separate frontend, so that the same backend could work with different frontends. A MediaServer module might take such role, taking some of the load off the shoulders of such projects and avoiding code duplication. The shared folders are a natural place to store media files, metadata and querying could be handled via filesystem indexing (tracker), epg could be centralized, capture cards could be in the server as opposed to the client to get rid of cables in the living room, lots of possibilities.... This would make it easier to penetrate the set-top-box market as soon as possible, which is important since we can then spread open codecs and be in a better position to fight DRM.
Finally there should be a small client-side application available also for windows and mac to register and configure such clients with the main server and install required open codecs.