Re: [IMRG] Requirements for IMP

[Maurizio Molina <molina@ccrle.nec.de>]

Hi Jon,
the scheme you propose is probably the one that gives the maximum 
flexibility with the minimum number of bits in the header (3).
However, in case you're intrested in measuring what happens on a set of 
X routers close to each other (i.e. on a sequence of next hops), AND 
these routers are far from the source (or on the reverse path) AND  you 
have a IPMP packet constrained in size,  you may have (in the worse 
case) to use X packets to let  each of these X routers fill the packet 
(playing with the TTL as you showed). Beyond having to spend X packets, 
you'll never be able to have in the same packet the timestamps of all 
these X neighbouring routers, which may be a limitation.
An example (numbers are only for the sake of clarifying what I mean): 
suppose your IPMP pkt can only contain 10 records and you're interested 
in measuring  
routers at "position" 21,22,23 along the path. Even if 3 path records 
would all fit in one IPMP message, with your mechanism you cannot 
specify N=0 (measure each hop),  Neither  N=1 (measure each two hops) as 
places for path records would be exhausted at router  10  (N=0) or at 
router 19 or 20 (N=1, depending if you set the initial TTL even or odd). 
To "reach" hop 21 you have to set N=2 (measure each 4 hops), but then no 
matters how to play with the TTL you'll only have simultaneously, into 
the IPMP pkt, only one of the records you're interested in.
To  select a range of routers I see no other way than spending 2 bytes 
(16 bits) in the header, with an explicit start and end index.
With 19 bits (start and end index + the 3 bits you suggested) we have 
probably all the flexibility we need.
I'm not pushing right now for one schema or the other, just wanted to 
point out that nothing comes for free...
Maurizio 
jon bennett wrote:

> Maurizio,
>
> when discussing measuring a sampling of routers you said
>
> >  or using a very few bits (an IPMP can write a paket only if the TLL 
> > ends with the same bit pattern).
>
> to do this you would need two things, you would need a mask of the bit 
> pattern, and a mask (or a count which would wind up being used by the 
> router to make a mask) of which bits out of the 8 in the TTL you 
> wanted to match against the pattern.
>
> in the interests of simplicity, would it suffice to have the packet be 
> stamped only if the TTL was some even multiple of 2? In other words 
> there would be for example the IMP packet would have a 3 bit field (N) 
> and only if the last N bits of the TTL were 1 (or 0?) would the packet 
> be stamped. this would let you test every 2nd, 4th, 8th, 16th, etc 
> router. By altering the initial TTL value you would be able to change 
> where the series started. for example if you had set N to 2 then a 
> packet sent with TTL=255 would measure
>
> TTL                     stamp   router
> 255=11111111            NO      1
> 254=11111110            NO      2
> 253=11111101            NO      3
> 252=11111100            YES     4
> 251=11111011            NO      5
> 250=11111010            NO      6
> 249=11111001            NO      7
> 248=11111000            YES     8
> etc
> yielding stamps by routers
> 4,8,12,16,20,etc
>
> but if instead you sent it with a TTL=253 you would get
>
> TTL                     stamp   router
> 253=11111101            NO      1
> 252=11111100            YES     2
> 251=11111011            NO      3
> 250=11111010            NO      4
> 249=11111001            NO      5
> 248=11111000            YES     6
> 247=11110111            NO      7
> 246=11110110            NO      8
> etc
> yielding stamps by routers
> 2,6,10,14,18,etc


_______________________________________________
IMRG mailing list
IMRG@ietf.org
https://www1.ietf.org/mailman/listinfo/imrg