[ih] Origin of "best effort"

[Brian E Carpenter]

On 21/01/2017 06:50, Michael Greenwald wrote:
> Are we asking about the specific phrase "best effort"?
> Or the notion that IP doesn't guarantee preserving
> packet sequence, or even guarantee delivery?

Specifically, "best effort" because that is not the same thing as
unreliable or out-of-order delivery (clearly mentioned in Pouzin's
early papers) or type-of-service networking (also in Pouzin and RFC791).

To be clear, packets delivered by a "best effort" mechanism and
packets delivered by a "type of service" or "differentiated services"
mechanism are all subject to loss or out-of-order delivery and therefore
all trigger the end to end argument.

As I said, Internet governance people mix up "best effort" with the
end to end argument, but the latter is concerned with *unreliable*
delivery, not with best effort specifically.

By the 1990s it was a commonplace that Internet service was best effort,
but that was by contrast with the unused TOS feature of RFC791.

    Brian

> 
> I assume not the latter, because the Catenet paper said:
> 
> "It is very important to note that it is explicitly assumed that
> datagrams are not necessarily kept in the same sequence on
> exiting a network as when they entered.  Furthermore, it is
> assumed that datagrams may be lost or even duplicated within the
> network.  It is left up to higher level protocols in the catenet
> model to recover from any problems these assumptions may
> introduce."
> 
> And RFC791 restates this as:
> 
> "The internet protocol does not provide a reliable communication
> facility.  There are no acknowledgments either end-to-end or
> hop-by-hop.  There is no error control for data, only a header
> checksum.  There are no retransmissions.  There is no flow control."
> 
> To my mind this explains what was later more loosely referred to
> as "best effort".  Is there any reason to believe it was used as
> a term of art, rather than a shorthand description of the 2 paragraphs
> I quoted? (I know that when I first heard "best effort" I
> interpreted it as a summary, but that just describes my reaction,
> not the intent of whoever used it).
> 
> On 2017-01-19 11:09, Guy Almes wrote:
>> Hi Brian,
>>    This is actually an interesting question.
>>    During the 1990s we had a discussion in which this phrase came up,
>> used in the now-typical networking sense.
>>    One colleague in the conversation, a lawyer, who had done work in 
>> the
>> construction industry, found the usage very odd because, he said, in
>> contracts within that industry the phrase had a specific meaning and it
>> obliged a person/company to a very very high standard of "best effort".
>> In that context, for example, it might oblige a company to spend
>> money/effort to a degree that would keep a promise but ruin any hope 
>> for
>> making money in the deal.  He noted that we computer engineers were
>> using the term in an almost opposite (and, to him, an ironic) sense,
>> viz., "do whatever is normal, but *not* heroic, and if it works, good
>> and if it doesn't, don't worry about it".
>>
>>    If the phrase is connected that older "contract language" usage, it
>> would indeed be interesting to see how it came to have almost a 
>> reversal
>> or sense.
>>
>> 	-- Guy
>>
>> On 1/18/17 2:34 PM, Brian E Carpenter wrote:
>>> I learnt very early on that the Internet offered a "best effort" 
>>> service
>>> for the delivery of datagrams.
>>>
>>> Where did that meme come from, and when?
>>>
>>> The earliest trace I found in a quick trawl was 1986 (RFC992). But 
>>> RFC791
>>> doesn't mention it, and defined TOS, such that all packets were *not* 
>>> assumed
>>> to be created equal. The 1984 Saltzer et al paper doesn't mention it 
>>> either.
>>>
>>> (RFC768 does say that UDP delivery is "not guaranteed" but that is not
>>> the same thing as "best effort".)
>>>
>>> The question is of interest because some analyses of network 
>>> neutrality,
>>> including a student dissertation I was reviewing yesterday, conflate 
>>> the
>>> end-to-end principle with best-effort packet delivery.
>>>
>>> Regards
>>>      Brian
>>> _______
>>> internet-history mailing list
>>> internet-history at postel.org
>>> http://mailman.postel.org/mailman/listinfo/internet-history
>>> Contact list-owner at postel.org for assistance.
>>>
>> _______
>> internet-history mailing list
>> internet-history at postel.org
>> http://mailman.postel.org/mailman/listinfo/internet-history
>> Contact list-owner at postel.org for assistance.
>