Experimental Physics and
Industrial Control System

"Brian McAllister" <[email protected]> · Fri, 15 Nov 2002 18:22:42 -0500

Kay-Uwe Kasemir wrote:

> I just think that writing files via NFS must be lame compared to
> ChannelAccess: NFS performs round-trip requests to "open", "write",
> "write", "write", ..  then "close".

Keep in mind that NFS writes in much larger pieces than CA does, so there
won't be that many "writes".

An example from one of our IOCs:

SR: 118 channels + header -> NFS file = 2853 bytes = 2 data packets

With the protocol overhead included, the difference in network load is
probably much less than you imply.

( Difficult to be quantitative about CA without that protocol spec.... )

- brian

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System