[foxboro] Nodebus Extender issues
- From: tom.vandewater@xxxxxxxxxxxxxx
- To: foxboro@xxxxxxxxxxxxx
- Date: Fri, 11 Oct 2002 17:02:22 -0400
List,
I am sending this because of several e-mails I have received off
list asking about the Nodebus Extender issues.
If you are expanding your installation, I would make every effort to
avoid using nodebus extenders. Creating another node and connecting the two
nodes with a carrierband would be more reliable, (albeit much more
expensive), than using nodebus extenders as they are today.
For any new expansions that require equipment to be seperated by a
distance that previously required NBE's, I would consider installing the new
redundant, solution that Foxboro is currently introducing. I don't know if
it is available already, but existing nodes can be interfaced to a
redundant, high speed network via the use of new nodebus modules called
NCNI's. Maybe someone from Foxboro can jump in here and tell us when we can
buy them. The NCNI's will allow Foxboro users that have nodebus
installations connected via the current Fox Carrierband 802.4, 5MB/Sec,
Token Bus, to migrate to a high speed switched ethernet running at either
100MB/Sec or 1GB/Sec.
I have heard talk of these same modules being used as a direct
replacement for NBE's and, technically it seems like it should work. We can
only hope that Foxboro corrected the NBE design problems with this new
offering!! They say they have and I think they will be using standard
firmware that is available in the communication market today. They will
just have to package it to fit in their nodebus form factor. This should be
the best solution for the replacement of the old NBE's. I think the big
delay in getting a fix for the NBE problem is because Foxboro doesn't want
to have to do a recall of all the NBE's in service and can't afford to
replace them with the new NCNI's!! I understand and appreciate their
dilemma.
The real problem with NBE's/FONBE's centers around their inherent
design issues and how they relay packets and handle collisions! Collisions
are more likely to occur because of the increased traffic rates that the
newer modules such as CP-60's and 51D's and above can generate. Foxboro
claims that the ethernet adapters in the D style boxes don't adher to the
IFG, (Inter Frame Gap), times that are specified in the 802.3 ethernet
standards and that causes problems for the NBE's when a D style box
communicates a lot of packets on the nodebus.
In addition, the newer modules, don't seem to wait as long, before
switching to the other nodebus when they find that the current bus is busy.
When a collision occurs, NBE's send out a jamming signal to force all
modules to cease communicating for a short period. The newer modules sense
that the bus is busy or failed when the NBE jams and they switch to the
other bus even though the bus they were on is not failed. When some
stations are on "A" and others are on "B" nodebus, Foxboro's dual bus
communication strategies fall apart. Even though there is an alternate path
available for traffic the stations are not smart enough to know which
stations are on which bus. Therefore, the stations communicating on "A"
send packets on "A" to stations that are only capable of responding on "B".
After multiple attempts and timeouts on "A" the station tries to communicate
on "B" and is successful, but the next time communication is required
between the same two stations, the same scenario occurs. The results are
massive slowdowns. This problem of stations communicating on opposite
busses also seems to create real problems with the FT strategies of CP's and
especially CBLANFT modules. During these disturbances, they begin to go
single.
The total problem is really much larger than NBE's. Foxboro's
Network Fault Detection, (NFD), strategies are flawed and need to be
revamped. I've heard they are working on NFD also. Because of all of the
problems currently existing with Foxboro's dual bus strategies, compounded
by the problems with NBE's, I have found that forcing one bus to stay
failed, is the only way I can keep nodes using NBE's and the newer style
nodebus stations, running without errors!
I learned all of this over the past year and a half, not because I
wanted to, but because I had to. Somebody from Foxboro that is more
knowledgeable than I, may want to correct or challenge my analysis of the
situation and I won't be offended if they tell me my interpretation of the
prolems is wrong. I would get extremely irate if they try to tell me that
there isn't really a problem.
Tom VandeWater
________________________________________________________________________
This email has been scanned for all viruses by the MessageLabs SkyScan
service. For more information on a proactive anti-virus service working
around the clock, around the globe, visit http://www.messagelabs.com
________________________________________________________________________
_______________________________________________________________________
This mailing list is neither sponsored nor endorsed by Invensys Process
Systems (formerly The Foxboro Company). Use the info you obtain here at
your own risks. Read http://www.thecassandraproject.org/disclaimer.html
foxboro mailing list: //www.freelists.org/list/foxboro
to subscribe: mailto:foxboro-request@xxxxxxxxxxxxx?subject=join
to unsubscribe: mailto:foxboro-request@xxxxxxxxxxxxx?subject=leave
Other related posts:
- » [foxboro] Nodebus Extender issues
