Post a Comment
Preparing to release?
have a look here:
http://syndicator.sun-catalogue.com/degermany/germanycontext" rel="nofollow">http://de.sun.com/catalog/?n-state=http://catalog.sun.com/productin...
you are a bit behind the times
Well for one, Citrix's ICA protocol is pretty high level. Does it even work with anything else than Windows (on the server, that is)?
One of the main advantages of Sun Ray is that the clients are almost completely dumb and have no moving parts. The protocol they use is very low level, so there's practically no processing client-side. Basically, they're just sending the framebuffer over Ethernet. IOW more like VNC, but IIRC Sun Ray is actually older than VNC.
Well for one, Citrix's ICA protocol is pretty high level. Does it even work with anything else than Windows (on the server, that is)?
Not quite correct; there is Citrix Metaframe which is the Winframe equivilant by on UNIX, hence the curiosity as to whether SUN licenced the protocol in Metaframe instead of developing their own transport protocol.
One of the main advantages of Sun Ray is that the clients are almost completely dumb and have no moving parts. The protocol they use is very low level, so there's practically no processing client-side. Basically, they're just sending the framebuffer over Ethernet. IOW more like VNC, but IIRC Sun Ray is actually older than VNC.
True, they probably use quite a lot of recycled technology from the old Java thin clients they sold a while back (they looked like a small expresso machine).
Nope. I've done remote X sessions over 25kB/s links with only a slight delay/lag. It really isn't bad at all. On even a 100mbit you'd be able to cram a crapload of these things. Considering any modern server is going to be spitting out data at 1gbit, dump that into a switch and pass it out as you see fit. I bet you could cram a thousand users on that bandwidth wise. You'd probably run into cpu/ram/etc issues far before you hit network capacity.
Nope. I've done remote X sessions over 25kB/s links with only a slight delay/lag. It really isn't bad at all. On even a 100mbit you'd be able to cram a crapload of these things. Considering any modern server is going to be spitting out data at 1gbit, dump that into a switch and pass it out as you see fit. I bet you could cram a thousand users on that bandwidth wise. You'd probably run into cpu/ram/etc issues far before you hit network capacity.
Going by the feedback I see on the SUN Blogs, there are end users who have SUN Ray appliances hooked up using a cable modem, so lets assume they compress it as well, end users have said that its like working on the network at SUN.
The problem as I see it, they need to drastically lower the cost of their appliancs, get it down to US$100-$150 per appliance; and heck, throw free SUN Ray appliances to customers who order the complete SUN Stack - both server and client end software.
Unfortunately, the SunRAYs use a proprietary protocol for booting and server connection. It isn't just a layer on top of X, but a framebuffer that gets (I assume) compressed and pushed over the network. The server software runs on Linux and Solaris, but isn't free in any way, and the Rays are useless without. It's especially bad because there isn't any use for the Rays once they get dumped--there was a box of them at the UCSB surplus a few months back, and I'd pick one up if only as a cheap music player or something, but no go.
I would say that it's the perfect moment to start a project to create an open source implementation of the SunRay server software (SRSS), at least with the basic functionalities.
It shouldn't be too difficult as there has already been some work done on trying to decode the protocol : see http://www.cl.cam.ac.uk/~mgk25/sunray/ or http://www.leonerd.org.uk/sunray/ for example.
At last, I could use my SunRay @ home without the actual software beast that SRSS is, which I find very difficult to install and to configure :-)
Anyone ready to tackle this challenge ?



