[ih] Fw: Flow Control in IP

[Barbara Denny]

just trying again to see if this will go through this time.

barbara
 
   ----- Forwarded Message ----- From: Barbara Denny <b_a_denny at yahoo.com>To: internet-history at elists.isoc.org <internet-history at elists.isoc.org>Sent: Tuesday, April 30, 2024 at 10:42:34 AM PDTSubject: Re: [ih] Flow Control in IP
  I don't know how much flow control at the IP level was discussed at the time. From the viewpoint of someone who worked on packet radio including flow control at the IP level would not be desirable.   
Some background, if  I remember correctly,  the ideal was to have the smallest  amount of state in the routers/gateways. Of course you needed routing and neighbor table information but I think that could be reconstructed rather quickly in case of failure. The model was a thin hourglass with IP at the waist. Packet radio and SATnet were the networks that were used in the original demonstrations of the Internet besides the ARPAnet. 
Packet radios could  dynamically adjust the bit rate  (either 100 or 400 kbits/second if my memory is correct) depending on conditions (link quality). The mac layer was csma-like (don't remember the details to map into all the flavors defined today).   When the radio could actually transmit (i.e broadcast ) depends on what the traffic is in your 2 hop neighborhood  to try to avoid a collision  due to hidden terminal problem(or potentially larger depending  on the capture capacity of the radio. You might be able to receive a packet and not decode it).  The packet radio would not try to transmit the next packet until it heard an ack for the packet if the next node was the destination  or it heard the next hop radio transmitting the packet on to its destination.  Also remember packet radio nodes were mobile.  

As I think you can see,  trying to do flow control across  a multihop packet radio network at the IP level might not work out very well. Unfortunately, the size of the deployed networks in testbeds and demonstrations as far as I know were small so I didn't get to do much experimenting here.  Rockwell was the developer of the radios and  implemented the protocols in the radio at this point in time.  I don't recall hearing how much work they did  in this area.  

barbara

    On Tuesday, April 30, 2024 at 12:41:04 AM PDT, Detlef Bosau via Internet-history <internet-history at elists.isoc.org> wrote:  
 
 Dear all,


may I ask this question?

RFC 791 explicitly states that there is NO flow control in IP.

As far as I see, some packet switching networks provide flow control
mechanisms, while others don't. So, I have some rough idea, why this
design decision was taken.

I wonder, whether there was a general discussion of this issue in the
past, particularly as this issue should have been relevant for the
ARPANET as well.

Many thanks for any comments on this one


Detlef