Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

Rate Control & far-end WIS (Re: PMD discussion)

Tom, Rich, Praveen,

I think we had better separate the local Rate Control from 
the far-end WIS implementation shown in Rich's slide #9.

In summary;
 (1) How does MAC recognize locally whether he is WAN-PHY or LAN-PHY?
      - I like to use station management (STA) register.

 (2) If we implement far-end WIS, do we need auto-negotiation?
      - No, but we will need 'local' auto-configuration.

Note that I have assumed that 802.3ae adopt the Shimon Muller's 
Open-Loop Control for the MAC rate down to 9.29 Gb/s.(Not the bit rate.)

(1)How does MAC recognize locally whether he is WAN-PHY or LAN-PHY?

Assuming that we have optional XAUI/XGXS, 10GbE router port 
could be implemented in separated two parts: a router/MAC 
package and an Attachement Unit (AU).  The latter may be a 
slot-in package or daughter card.  In this case, MAC should 
recognize whether his AU is LAN-PHY or WAN-PHY.  The STA 
register would be the promising candidate for this purpose.  
I don't think manual setting at each AU installation is less 
troublesome than defining/using a register bit for this auto 

In my sense, this auto configuration is far from 'negotiation'. 
MAC will check the register bit when he recognize that the 
AU is newly installed.  No timing issue.  No acknowledgement.  
I think this STA register bit is extremely useful and worth 
for the standard.

(2) If we implement far-end WIS, do we need auto-negotiation?

No, I don't think so.  We can use similar 'auto-configuration' 
mechanism.   LSS provides the Layer-1 information reporting 
from far-end STA to local STA, so it could send the far-end 
'WIS' existence or a register bit to the local STA periodically 
by using Link Status [r] together with Remote Fault and Break Link.
I don't think this is WAN mode 'negotiation'; it's still just a 
reporting mechanism.  No timing/speed issue here; always 
advertising the same bit status.

In this far-end WIS implementation,  the local MAC should recognize 
the 'far-end AU/PHY installation' event to initiate the auto-
configuration of his local MAC rate.  Every time the Link is 
initialized, MAC had better check his local STA register bit 
that is reflecting the far-end WIS existence.

In short, if the community reaches the consensus not to prohibit 
anyone from implementing far-end WIS, LSS could provide the 
practical solution without adding any control codes.

Best Regards,

At 03:59 00/06/03 -0700, Richard Taborek wrote:
> Tom Alexander wrote:
> >   - the 802.3ae objectives do not provide for any form of
> >     autonegotiation over the link. The proposed LSS does
> >     not include speed negotiations, nor is it likely to
> >     be able to do so with current coding schemes.
> WAN mode would be configured by SMT in the same manner as other 
> options such as flow control, etc.
> The primary reason for the committees rejection of Auto-Negotiation 
> at 10 Gbps is that the preferred mode of link management is to 
> configure a link and have the link come up in the same mode after a 
> power event or other disturbance.  There is no apparent benefit in 
> having a 10 GbE backbone negotiate down to 1 Gbps on its own accord 
> like a modem. 
> LSS can easily accommodate WAN mode negotiation. I don't believe that 
> we have any speeds to negotiate.

At 19:19 00/06/02 -0700, Praveen Kumar wrote:
> In summary, I don't see why parking the WIS on the far end will not 
> work.  That being said, Iam not sure either, about whether there 
> are enough benefits in allowing this to be a part of the standard 
> (it sure does cause some confusion).

Osamu Ishida,ishida@xxxxxxxxxxxxxxxxxxx
NTT Network Innovation Laboratories
TEL:+81-468-59-3263 FAX:+81-468-55-1282