Re: [mif] Five additional problem statements for mif ---- host routing

[Min Hui <[e-mail]>]

Hi, Lars

Thanks for your clarify. I also agree that app can bind a source
address to solve this problem, but it will be better if the stack can
do this.
I think we reach a consensus that there exists the problem of host
routing regardless whether the stack or app should be responsible for
this.
So would you please let me know what's your suggestion about the
description of this problem?

Thanks.

-Hui Min

2009/3/18 Lars Eggert <[e-mail]>:
> Hi,
>
> well, you could also argue that the current source address selection is
> actually fine, and it's in fact the applications' fault, because they don't
> bind to a specific address. Not binding to an address is telling the OS "you
> choose for me". As long as there is only one interface, being lazy has no
> consequences, but if there are multiple interfaces, it does.
>
> I agree that fixing this in the stack has distinct advantages over waiting
> for apps to be fixed, but I disagree that the current stack mechanism is
> broken - it isn't. It's the apps that are mostly broken.
>
> Lars
>
> On 2009-3-17, at 20:45, Min Hui wrote:
>
>> Hi, everyone
>>
>> This is the first one: Host routing
>>
>> The host routing currently follows the default gateway mechanism,
>> which will choose the unify gateway among more than one default routes
>> ('0.0.0.0'), the detail is described in RFC1122. The default gateway
>> guarantees there always has a route to network when the host can not
>> find a specific route for a datagram in the route table.
>> But when it comes to multiple interfaces situation, the default
>> gateway mechanism in host routing will let all the IP flows go out
>> through one interface except some specific assignment (e.g. static
>> routing item). In this case, the applications can’t use different
>> interfaces which are the aim of multiple interfaces. The reason is the
>> application will not appoint a source address in most situations
>> currently. The source address will be determined by host operating
>> system after querying the host routing table, if there is any
>> available routing item for the destination, the corresponding source
>> address of this routing item will be selected. In most cases, the
>> routing item of default gateway will be used, so that every
>> application will use the same interface.
>> In conclusion, the above host routing mechanism is one of problems in
>> the way to maintain multiple interfaces work simultaneously.
>>
>> - Hui Min
>>
>> 2009/3/18 Min Hui <[e-mail]>:
>>>
>>> Hi, all mif fans
>>>
>>> I have viewed the agenda of mif bof, and I notice some problem
>>> statements for mif are not concluded in the current ps documents which
>>> will be discussed in the meeting.
>>> So I post these additional problems in the mail list for discussion,
>>> all of them come from the draft
>>> "draft-hui-ip-multiple-connections-ps-02" with some modifications
>>> according to the comments received recently.
>>> There are five problems in addition, which will be proposed in five
>>> separated mails in order to have sufficient discussion for each of
>>> them.
>>> Five mails will be sent out following this mail, the structure is:
>>> 1. Host routing
>>> 2. DNS selection
>>> 3. Different metric measurements
>>> 4. Source address selection for IPv4
>>> 5. TOS consideration
>>>
>>> Any comment is welcomed, and thanks for your notice.
>>>
>>> BR,
>>> - Hui Min
>>>
>> _______________________________________________
>> mif mailing list
>> [e-mail]
>> https://www.ietf.org/mailman/listinfo/mif
>
>