[Pipet Devel] Re: which handles distribution (was: vsh?)
jarl van katwijk
jarl at casema.net
Fri Mar 17 15:41:27 EST 2000
>
> This whole 'speed' issue has me confused. If the small Overflow objects
> remain 'networked' on a sigle host (none are sent through the Internet), how
> can GMS affect their speed???
Every data flow will go thru
a sensor ->
-> the pipenline ->
-> collector (can contain a application, a internal procedure or a
overflow subnet) ->
-> visual (GUI or whatever).
This currently goes pretty slow. Lots of IIOP traffic and memory access. I'm
currently coding
a benchmark plugin.. more on this later.
NOTE that the collector can also read it's own data from disk, open local
sockets, etc
BUT it will be authenticated\controlled\bla-bla-bla.
>
>
> And if small Overflow objects are 'networked' across the Internet, IT'S THE
> INTERNET THAT WILL SLOW THEM DOWN, NOT A GMS WRAPPER. Isn't that obvious to
> everone else?
Hehe.. I think the concept of just using the overflow library for subnets caused
some confusion :)
>
> As I mentioned before, I'd like to see the GUI communicate with GMS, GMS
> communicate with Overflow across the Internet, and Overflow handle application
> wrapping/binding and development. So, each and every GMS node is an instance
> of Overflow on the Internet, which may in many cases connect to foreign
> applications running on that host.
>
> +------------+
> | Loci GUI |
> +------------+
> |
> |
> |
> |
> \|/
> +-----------+
> | GMS |
> +-----------+
> / \
> / \
> INTERNET / \ INTERNET
> / \
> / \
> | |
> \|/ \|/
> +-----------+ +-----------+
> | GMS | | GMS |
> +-----------+ +-----------+
> | |
> | |
> \|/ \|/
> +-----------+ +-----------+
> | Overflow | | Overflow |
> +-----------+ +-----------+
> | |
> | |
> \|/ \|/
> +-------------+ +-------------+
> | Application | | Application |
> +-------------+ +-------------+
>
Fine. I will erase 'Application wrapping' of my TODO..
why was there so much email needed for this concept 8)
More information about the Pipet-Devel
mailing list