Modbus TCP or NDDS

M

Thread Starter

Mark Meng

I am rather confused by reading in the press release that there are two communication/fieldbus technologies on Fast Ethernet: ModbusTCP and NDDS, both supported by initiated by Schneider. Everybody knows ModbusTCP, but it there any product on NDDS launched? Will it replace ModbucTCP in future since it is said that NDDS has so many advantages?
 
Dear Sir:

Modbus TCP/IP and RTPS/NDDS are complementary technologies (protocols) and they can coexist in the same communication stack with no problem at all, in the same way today Modbus TCP/IP coexists today with multiple Internet communication protocols. This is one of the benefits Schneider provides to Transparent Factory users: the possibility to have the right protocol for the right application. Both, RTPS/NDDS (publisher/subscriber) and Modbus TCP/IP (client/server messaging) are oriented not only to different applications but also in some cases to different audiences.

 
A

Armin Steinhoff

> I am rather confused by reading in the press release that there are two communication/fieldbus technologies on Fast Ethernet: ModbusTCP and NDDS, both supported by initiated by Schneider.<

NDDS is a proprietary protocol (IP multicast based) from RTI ... it is not as open specified as the ModbusTCP protocol.

I don't know whether Schneider will use this
product from RTI or not.

> Everybody knows ModbusTCP, but it there any >product on NDDS launched? Will it replace >ModbucTCP in future since it is said that NDDS >has so many advantages? <

I'm not convinced ... NDDS doesn't care about
QoS aspects.

If you are interested in reliable multicast protocols ... visit:

http://lgmp.planethofmann.de -> Multicast Links
-> Software


Regards

Armin Steinhoff


 
P

Peter Bolton

> > I am rather confused by reading in the press release that there are two communication/fieldbus technologies on Fast Ethernet: ModbusTCP and NDDS, both supported by initiated by Schneider.<
>
> NDDS is a proprietary protocol (IP multicast based) from RTI ... it is not as open specified as the ModbusTCP protocol. <clip><

NDDS is an implementation of a formal real-time publish-subscribe (RTPS) protocol. Real-Time Innovations (RTI) will publish the protocol specification so others can implement it. A draft of the RTPS protocol specification will be included with the Interface For Distributed Automation specification to be available at the SPS/IPC show in November. The full RTPS specification alone is planned for publication later this year as an IETF Informational RFC.

For more information about the Interface for Distributed Automation Group, go to www.ida-group.org.

If you would like to be notified when the full RTPS specification is available send an email to [email protected]
 
L

Lynn August Linse

>How will Schneider position these two totally incompatible technologies
>based on Fast Ethernet?
>
>I am rather confused by reading in the press release that there are two
>communication/fieldbus technologies on Fast Ethernet: ModbusTCP and
>NDDS, both supported by initiated by Schneider. Everybody knows
>ModbusTCP, but it ther any product on NDDS launched? Will it replace
>ModbucTCP in future since it is said that NDDS has so many advantages?

They are really apples & oranges - depending on how you view the situation at the time. NDDS exists because multi-cast is critical to future Ethernet based systems & Modbus/TCP as-is doesn't support multi-cast. It could be expanded easily to do it, but possibly some people felt Modbus/TCP was "too open" & made it difficult to create brand advantage.

Three Example Views:

1) Time to implement (without buying outside code)
- Modbus/TCP a few days verse NDDS a few months or years

2) Network load
- Modbus/TCP with explicit poll-request, likely at least 4 times MORE traffic than NDDS, but in large systems this could 100's of time more traffic for the same effect

3) Code size & HW resources
- Modbus/TCP is probably one of the smallest, NDDS will be considerably more

Other Q-n-A:

Will NDDS replace Modbus/TCP?
- for single-vendor Schneider or Connect-partner systems, likely yes.
- for 3rd party or multi-vendor systems, likely NO

Will Rockwell, Siemens, GE or others support NDDS?
- a snow-ball's chance in h*ll

Will NDDS become a world-wide standard?
- No chance. There are too many secrets within it. Too many things where the players say "it's open" and when you ask for details they say "the function's open, so you're free to use your own function, we use our own & won't tell you about that, but since you can use your own function it's open ..." This kind of "open standard" falls under the old adage "You can fool some of the people some of the time ... but you cannot fool all of the people all of the time."


Best Regards

Lynn August Linse, [email protected] http://www.linse.org/lynn
3 Rue Monet, Foothill Ranch CA 92610
Ph: 949-300-6337 Fx: 612-677-3253


 
Thanks, Lynn, you answered more than I asked. Which Industrial Ethernet protocal do you think will gain more acceptance in the near future?
 
T
Modbus segments seems likely to become a good short term solution in applications not requiring 'clockwork' determinism (really implying that the devices intelligently do their job, for example PID Controllers, Drives,
Process Apps, but not distributed I/O). Obvious problem is the lack of standardisation of data semantics, but it isn't a massively big issue, or at least needn't be. I'd certainly be betting in this direction short to medium term.

For distributed I/O applications, I guess you'll see either continuation of traditional fieldbuses like Profibus/CanOpen,DeviceNet, or possibly
clustered I/O controllers on Ethernet capable of providing some sort of time synchronised protocol (which hasn't really been invented yet). You need an I/O cluster to bear the additional cost of the Ethernet hardware (including a capable microprocessor, etc) but really I wouldn't see what benefits using Ethernet would give over existing fieldbuses. Similarly, I really can't see
why I'd want to run Profibus over Ethernet (for example), as it really just adds complications to a fairly optimum solution.

At present - hype apart (which seems to be diminishing a little) I can't see much driving adoption of Ethernet down into the device level. ControlNet, which was my great hope, requires a highly capable processor and RTOS (VXWorks type with a large number of tasks). The cost benefits of using Ethernet are pretty dubious when you really look at them (factoring in the cost of industrial grade switches, cable and so on). Data transparency right across an enterprise network is in some senses never going to be achievable
using Ethernet - you need private segments to guarantee response - and in any case possible via a data server/gateway using any sub-network
technology. I also think some of the initial suggestions for Ethernet apps, such as Web config and E-Mailed alarms are really nothing more than toy functions.

Tim Linnell (Eurotherm Controls - but all opinions are my own!).
 
Top