User:Sancelot
From RTwiki
(Difference between revisions)
| Line 1: | Line 1: | ||
realtime software developer since 90's | realtime software developer since 90's | ||
| + | |||
| + | = INDUSTRIAL REALTIME ETHERNET = | ||
| + | |||
| + | Linux preempt RT seems a good choice for developing realtime internet networks | ||
| + | |||
| + | Well known name of the moment : | ||
| + | * profinet | ||
| + | * powerlink | ||
| + | * ethercat | ||
| + | * modbus tcp | ||
| + | * ethernet/ip | ||
| + | ..... | ||
| + | gateways to older well known fieldbuses : canopen, ... | ||
| + | |||
| + | I will not tell if one is better or not than another, this completely depends on YOUR requirements. | ||
| + | |||
| + | The today's problem is that we will need a combination of all for some reasons... (retrofitting applications, migrating to new protocols , availability of products ....) | ||
| + | |||
| + | This is why we would need to develop : | ||
| + | The '''Opensource realtime network industrial framework''' , a software gateway running linux rt preempt able to dialog with any bus (selected at compilation). | ||
| + | using standard Ethernet hardware. | ||
| + | |||
| + | I encourage everybody beginning the discussion here , to create the project . | ||
| + | |||
| + | a quick TODO list , contributors welcomed , I know some people are already on the way : | ||
| + | * measurement and publication of timing spent to transverse the linux kernel network stack needed in order finding where optimisations are needed. | ||
| + | * searching for the best software architecture in order being able driving ethernet protocols (ace api ? ,other ...) | ||
| + | * optimisations of network drivers : most ethernet linux drivers do not provide the ability to use hardware features available in the chipsets. (eg UPD checksum ..) | ||
Revision as of 09:11, 28 October 2008
realtime software developer since 90's
INDUSTRIAL REALTIME ETHERNET
Linux preempt RT seems a good choice for developing realtime internet networks
Well known name of the moment :
- profinet
- powerlink
- ethercat
- modbus tcp
- ethernet/ip
..... gateways to older well known fieldbuses : canopen, ...
I will not tell if one is better or not than another, this completely depends on YOUR requirements.
The today's problem is that we will need a combination of all for some reasons... (retrofitting applications, migrating to new protocols , availability of products ....)
This is why we would need to develop : The Opensource realtime network industrial framework , a software gateway running linux rt preempt able to dialog with any bus (selected at compilation). using standard Ethernet hardware.
I encourage everybody beginning the discussion here , to create the project .
a quick TODO list , contributors welcomed , I know some people are already on the way :
- measurement and publication of timing spent to transverse the linux kernel network stack needed in order finding where optimisations are needed.
- searching for the best software architecture in order being able driving ethernet protocols (ace api ? ,other ...)
- optimisations of network drivers : most ethernet linux drivers do not provide the ability to use hardware features available in the chipsets. (eg UPD checksum ..)
