Experimental Physics and
Industrial Control System

Andrew Johnson <[email protected]> · Fri, 24 Sep 2010 12:10:08 -0500

On Friday 24 September 2010 11:24:42 Ralph Lange wrote:
>   On 24.09.2010 12:16, Eric Norum wrote:
> > Or run every IOC in it's own virtual machine?
>
> Have been investigating that- too thick. Running a VM only to have an
> additional IP address .... nah.
>
> I would prefer the solution that uses virtual network interfaces on the
> soft IOC host. Relatively slim, completely transparent, does not need
> additional services.

But it does require configuring a alias interface and providing an IP address 
for each soft IOC you want to run.  If you're using non-routable addresses the 
extra IPs may not be a major issue (as long as your subnets were made large 
enough), but it might mean getting IT involved every time Controls wants to 
add a new soft IOC.  Do you have to configure CA clients running on a machine 
with multiple interfaces to only use one of them for broadcasts, or would CA 
take care of that for you if they're all in the same subnet?

A machine-specific name-server would be more efficient since each PV name 
request only has to be checked once by the name-server rather than by all of 
the IOCs on the machine.

- Andrew
-- 
The best FOSS code is written to be read by other humans -- Harald Welte

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System