Loading ...
Sorry, an error occurred while loading the content.

CLB-013 - Network Port Object

Expand Messages
  • Carl Neilson
    BACneteers, I have received a few comments on the Network Port object from a developer at Delta Controls and would like to share those comments before
    Message 1 of 5 , May 25, 2011
    • 0 Attachment

      BACneteers,

       

      I have received a few comments on the Network Port object from a developer at Delta Controls and would like to share those comments before Montreal:

       

      - NetworkPort.MAC_Address is defined as type BACnetMACAddress.  The BACnetMACAddress is defined as a Sequence with only one field, a mac-address octet string.  It would be cleaner if NetworkPort.MAC_Address were defined as an octet string type directly, skipping the extra sequence tagging.

       

      - NetworkPort.BACnet_IP_Subnet_Mask is defined as type BACnetHostAddress.  An IPv4 subnet mask can never be specified by DNS hostname so a base octet string would be a more appropriate type than BACnetHostAddress.

       

      - NetworkPort.BACnet_IP_Default_Gateway is defined as type BACnetHostAddress.  A default gateway isn’t defined by DNS hostname so a base octet string would probably be more appropriate.

       

      - Dev.Home-Port – It is not clear why devices have to be single homed.

       

      There has been a comment in CLB-013 since revision 11 indicating that the need for a Home network be reviewed with respect to the requirements for Network Security and that if it is to remain in the document that a definition of a home port be added. At this point I am not convinced that the standard, including network security, requires single homed devices. This should be evaluated before Montreal and removed if we cannot find the requirement in the existing standard.

       

      Carl

    • Coleman Brumley
      Carl, Thanks for the feedback. The first 3 points are cut and dried and will be addressed prior to Montreal. As far as the 4th point... I loathe to re-open
      Message 2 of 5 , May 26, 2011
      • 0 Attachment

        Carl,

         

        Thanks for the feedback.  The first 3 points are cut and dried and will be addressed prior to Montreal. 

         

        As far as the 4th point...

         

        I loathe to re-open this can of worms because it's going to be a big discussion no matter how we slice it.

         

        The following is from an email from David Fisher on 7-Sep-2010.  I believe this points to the requirement of a single home for devices in clause 5.1. 

         

        For those of you wanting to know where the standard defines this binding idea it is

        described in detail at the end of 5.1 (since the first 1995 standard):

        A "BACnetDevice" is any device, real or virtual, that supports digital communication

        using the BACnet protocol. Each BACnet Device contains exactly one Device Object,

        as defined in 12.10. A BACnet Device is uniquely located by an NSAP, which consists

        of a network number and a MAC address.

         

        But, as Bill Swan points out on 20-Jan-2011:

         

        The phrase says “located” and not “homed.”  A multi-homed router can still be uniquely

        “located” (found) at a single address by some other device, no matter where that

        device is in the system.  Two devices in different locations have different NSAPs for

        the target device in their tables, but the system still works.

         

        As you pointed out, also on 20-Jan-2011:

         

        In post meeting discussions, Bill and I agreed that the issue has little impact on

        real interoperability as shown by numerous real-world installations that show no ill effects

         

        I agree with your point about interoperability.  It shouldn't matter one way or another to the rest of the network whether the binding looks like it's a routed address or not -- assuming the routing information is correct. 

         

        Frankly, I don't care either way and I really don't want to hold this proposal up because of this ages long argument. 

         

        So, how about a "straw poll" for one of the following 2 options:

         

        1) The property is marked as required currently.  How about if we just make it optional?  Then those who use the "single homed" approach can  use it, and those who use the "multi homed" approach won't. 

        2) Remove it entirely.  I can make an argument for this, because how is it used anyway?  What's the use case for this property? 

         

        - Coleman

         

        From: bacnet-oswg@yahoogroups.com [mailto:bacnet-oswg@yahoogroups.com] On Behalf Of Carl Neilson
        Sent: Wednesday, May 25, 2011 6:23 PM
        To: bacnet-oswg@yahoogroups.com; bacnet-ip-wg@yahoogroups.com
        Subject: [bacnet-oswg] CLB-013 - Network Port Object

         

         

        BACneteers,

         

        I have received a few comments on the Network Port object from a developer at Delta Controls and would like to share those comments before Montreal:

         

        - NetworkPort.MAC_Address is defined as type BACnetMACAddress.  The BACnetMACAddress is defined as a Sequence with only one field, a mac-address octet string.  It would be cleaner if NetworkPort.MAC_Address were defined as an octet string type directly, skipping the extra sequence tagging.

         

        - NetworkPort.BACnet_IP_Subnet_Mask is defined as type BACnetHostAddress.  An IPv4 subnet mask can never be specified by DNS hostname so a base octet string would be a more appropriate type than BACnetHostAddress.

         

        - NetworkPort.BACnet_IP_Default_Gateway is defined as type BACnetHostAddress.  A default gateway isn’t defined by DNS hostname so a base octet string would probably be more appropriate.

         

        - Dev.Home-Port – It is not clear why devices have to be single homed.

         

        There has been a comment in CLB-013 since revision 11 indicating that the need for a Home network be reviewed with respect to the requirements for Network Security and that if it is to remain in the document that a definition of a home port be added. At this point I am not convinced that the standard, including network security, requires single homed devices. This should be evaluated before Montreal and removed if we cannot find the requirement in the existing standard.

         

        Carl

      • deanmatsen
        Regarding single-homed devices, I always thought it was optional to be a single-homed device, but this just gives us a way to describe the NetworkPort object
        Message 3 of 5 , May 26, 2011
        • 0 Attachment
          Regarding single-homed devices, I always thought it was optional to be a single-homed device, but this just gives us a way to describe the NetworkPort object for such devices.

          Last time I looked at the NetworkPort object, it seemed like the idea was that single-homed implementations would have the property, and other devices wouldn't.

          My conclusion is that secure devices pretty much do need to be single-homed, because of the relationship between the device base security and the individual network policies. I also think single-homing solves some problems with NAT.

          If nothing else, we need to keep the idea of single-homed implementations available as an option.

          --- In bacnet-ip-wg@yahoogroups.com, "Carl Neilson" <cneilson@...> wrote:
          >
          > BACneteers,
          >
          >
          >
          > I have received a few comments on the Network Port object from a
          > developer at Delta Controls and would like to share those comments
          > before Montreal:
          >
          >
          >
          > - NetworkPort.MAC_Address is defined as type BACnetMACAddress. The
          > BACnetMACAddress is defined as a Sequence with only one field, a
          > mac-address octet string. It would be cleaner if
          > NetworkPort.MAC_Address were defined as an octet string type directly,
          > skipping the extra sequence tagging.
          >
          >
          >
          > - NetworkPort.BACnet_IP_Subnet_Mask is defined as type
          > BACnetHostAddress. An IPv4 subnet mask can never be specified by DNS
          > hostname so a base octet string would be a more appropriate type than
          > BACnetHostAddress.
          >
          >
          >
          > - NetworkPort.BACnet_IP_Default_Gateway is defined as type
          > BACnetHostAddress. A default gateway isn't defined by DNS hostname so a
          > base octet string would probably be more appropriate.
          >
          >
          >
          > - Dev.Home-Port - It is not clear why devices have to be single homed.
          >
          >
          >
          > There has been a comment in CLB-013 since revision 11 indicating that
          > the need for a Home network be reviewed with respect to the requirements
          > for Network Security and that if it is to remain in the document that a
          > definition of a home port be added. At this point I am not convinced
          > that the standard, including network security, requires single homed
          > devices. This should be evaluated before Montreal and removed if we
          > cannot find the requirement in the existing standard.
          >
          >
          >
          > Carl
          >
        • Coleman Brumley
          Dean, I don t think removing the property necessarily removes the option of being a single homed device. It just removes the BACnet application visible aspect
          Message 4 of 5 , May 27, 2011
          • 0 Attachment

            Dean,

             

            I don't think removing the property necessarily removes the option of being a single homed device.  It just removes the BACnet application visible aspect of it.  Meaning, that if we remove the property, then there's no way to tell via ReadProperty (for example) if a device is single homed or not.  I'm not certain that's a problem or not, and that's the only use case I envision for that property.  There may be others, especially regarding NS, but I'm having a hard time coming up with anything else. 

             

            Coleman

             

            From: bacnet-ip-wg@yahoogroups.com [mailto:bacnet-ip-wg@yahoogroups.com] On Behalf Of deanmatsen
            Sent: Thursday, May 26, 2011 9:54 PM
            To: bacnet-ip-wg@yahoogroups.com
            Subject: [bacnet-ip-wg] Re: CLB-013 - Network Port Object

             

             

            Regarding single-homed devices, I always thought it was optional to be a single-homed device, but this just gives us a way to describe the NetworkPort object for such devices.

            Last time I looked at the NetworkPort object, it seemed like the idea was that single-homed implementations would have the property, and other devices wouldn't.

            My conclusion is that secure devices pretty much do need to be single-homed, because of the relationship between the device base security and the individual network policies. I also think single-homing solves some problems with NAT.

            If nothing else, we need to keep the idea of single-homed implementations available as an option.

            --- In bacnet-ip-wg@yahoogroups.com, "Carl Neilson" <cneilson@...> wrote:
            >
            > BACneteers,
            >
            >
            >
            > I have received a few comments on the Network Port object from a
            > developer at Delta Controls and would like to share those comments
            > before Montreal:
            >
            >
            >
            > - NetworkPort.MAC_Address is defined as type BACnetMACAddress. The
            > BACnetMACAddress is defined as a Sequence with only one field, a
            > mac-address octet string. It would be cleaner if
            > NetworkPort.MAC_Address were defined as an octet string type directly,
            > skipping the extra sequence tagging.
            >
            >
            >
            > - NetworkPort.BACnet_IP_Subnet_Mask is defined as type
            > BACnetHostAddress. An IPv4 subnet mask can never be specified by DNS
            > hostname so a base octet string would be a more appropriate type than
            > BACnetHostAddress.
            >
            >
            >
            > - NetworkPort.BACnet_IP_Default_Gateway is defined as type
            > BACnetHostAddress. A default gateway isn't defined by DNS hostname so a
            > base octet string would probably be more appropriate.
            >
            >
            >
            > - Dev.Home-Port - It is not clear why devices have to be single homed.
            >
            >
            >
            > There has been a comment in CLB-013 since revision 11 indicating that
            > the need for a Home network be reviewed with respect to the requirements
            > for Network Security and that if it is to remain in the document that a
            > definition of a home port be added. At this point I am not convinced
            > that the standard, including network security, requires single homed
            > devices. This should be evaluated before Montreal and removed if we
            > cannot find the requirement in the existing standard.
            >
            >
            >
            > Carl
            >

          • Matsen, Dean C
            I understand what you mean. But that s assuming the device is able to automatically choose its home port in all cases. I think the ultimate benefit of the
            Message 5 of 5 , May 27, 2011
            • 0 Attachment

              I understand what you mean.  But that's assuming the device is able to automatically choose its home port in all cases.

               

              I think the ultimate benefit of the NetworkPort object is that anyone can configure anyone else's devices.  If we leave this concept out of the NetworkPort object, then control of the home port will become hidden in the "local matter" realm and kind of defeat the purpose.  Since single-homing is a current practice with some vendors, a standard way of controlling it should be included in the NetworkPort object.  

               

              I'm not saying we should overly specify the behavior of this property either.  I wouldn't want to force all vendors that do single-homing to allow the property to be writable.  It could be read-only and informational in some vendor's devices.  I do think the presence or absence of the property would be a good way to determine if a device is a single- homed implementation or not.

               

              Dean Matsen

              Engineer Software Pr

              Alerton Dealer Business

              Honeywell Automation & Control Solutions

              6670 185th Ave NE

              Redmond WA 98052

              Phone – 425.897.3980

              Fax – 425.869.8445

              dean.matsen@...

               


              From: bacnet-ip-wg@yahoogroups.com [mailto:bacnet-ip-wg@yahoogroups.com] On Behalf Of Coleman Brumley
              Sent: Friday, May 27, 2011 5:44 AM
              To: bacnet-ip-wg@yahoogroups.com
              Subject: RE: [bacnet-ip-wg] Re: CLB-013 - Network Port Object

               

               

              Dean,

               

              I don't think removing the property necessarily removes the option of being a single homed device.  It just removes the BACnet application visible aspect of it.  Meaning, that if we remove the property, then there's no way to tell via ReadProperty (for example) if a device is single homed or not.  I'm not certain that's a problem or not, and that's the only use case I envision for that property.  There may be others, especially regarding NS, but I'm having a hard time coming up with anything else. 

               

              Coleman

               

              From: bacnet-ip-wg@yahoogroups.com [mailto:bacnet-ip-wg@yahoogroups.com] On Behalf Of deanmatsen
              Sent: Thursday, May 26, 2011 9:54 PM
              To: bacnet-ip-wg@yahoogroups.com
              Subject: [bacnet-ip-wg] Re: CLB-013 - Network Port Object

               

               

              Regarding single-homed devices, I always thought it was optional to be a single-homed device, but this just gives us a way to describe the NetworkPort object for such devices.

              Last time I looked at the NetworkPort object, it seemed like the idea was that single-homed implementations would have the property, and other devices wouldn't.

              My conclusion is that secure devices pretty much do need to be single-homed, because of the relationship between the device base security and the individual network policies. I also think single-homing solves some problems with NAT.

              If nothing else, we need to keep the idea of single-homed implementations available as an option.

              --- In bacnet-ip-wg@yahoogroups.com, "Carl Neilson" <cneilson@...> wrote:

              >
              > BACneteers,
              >
              >
              >
              > I have received a few comments on the Network Port object from a
              > developer at Delta Controls and would like to share those comments
              > before Montreal:
              >
              >
              >
              > - NetworkPort.MAC_Address is defined as type BACnetMACAddress. The
              > BACnetMACAddress is defined as a Sequence with only one field, a
              > mac-address octet string. It would be cleaner if
              > NetworkPort.MAC_Address were defined as an octet string type directly,
              > skipping the extra sequence tagging.
              >
              >
              >
              > - NetworkPort.BACnet_IP_Subnet_Mask is defined as type
              > BACnetHostAddress. An IPv4 subnet mask can never be specified by DNS
              > hostname so a base octet string would be a more appropriate type than
              > BACnetHostAddress.
              >
              >
              >
              > - NetworkPort.BACnet_IP_Default_Gateway is defined as type
              > BACnetHostAddress. A default gateway isn't defined by DNS hostname so a
              > base octet string would probably be more appropriate.
              >
              >
              >
              > - Dev.Home-Port - It is not clear why devices have to be single homed.
              >
              >
              >
              > There has been a comment in CLB-013 since revision 11 indicating that
              > the need for a Home network be reviewed with respect to the requirements
              > for Network Security and that if it is to remain in the document that a
              > definition of a home port be added. At this point I am not convinced
              > that the standard, including network security, requires single homed
              > devices. This should be evaluated before Montreal and removed if we
              > cannot find the requirement in the existing standard.
              >
              >
              >
              > Carl
              >

            Your message has been successfully submitted and would be delivered to recipients shortly.