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

RE: Link Status encodings




Rich,

There are fairly few control codes available in the 10B space. Before we
decide to burn another one for ordered sets, I would like to see Fibre
Channel figure out whether they can use the "Pulse" style insertion in place
of the sequence ordered sets. As I said at the meeting, it is not obvious
that the sequence insertion that you have proposed to Fibre Channel yields
acceptable EMI spectrum. We could divide the code space of the pulse ordered
sets between Ethernet and FC usage. For instance, we could give half the
values of the first data byte to each.

Also, I agree with Birdy that the name Pulse Ordered-Set does not denote its
operation and is confusing because of the normal physical meaning of pulse.
Other possible candidates:

  Phy Ordered Set
  Link Ordered Set

Pat

-----Original Message-----
From: Rich Taborek [mailto:rtaborek@xxxxxxxxxxxxx]
Sent: Friday, November 17, 2000 11:42 AM
To: HSSG
Subject: Re: Link Status encodings



Birdy,

I called the Ordered-Set Pulse instead of Link Status because Link
Status represents only the two messages defined for transport, namely
Local Fault and Remote Fault. A third message Break Link, is proposed
but not accepted. It is not a Link Status, but rather a Link Request. In
the future 10GE may have a use for Pulse Ordered-Sets for some reason
perhaps having nothing to go with Link Status. I wanted to chose a
generic name to prevent a potential future naming conflict.

As far a running disparity goes, I see a potential desire for 10GE
implementations to replace an ||R|| with a ||P|| at the 10b level. This
would not be possible if I chose a non-neutral disparity special
code-group for the /K/ of ||P||. All data code-groups associated with
||P|| would need to observe the same rules. However, we're quickly using
up the 12 special code-groups and the /D/ codes are much easier to
change down the line if required. Yes, it's half-baked, but it provides
flexibility and affects nothing. You are 100% correct none of this
discussion is applicable to ||P|| inserted at the 8b level.

In November a liaison letter was forwarded from the chair of NCITS T11
to the chair of IEEE P802.3ae requesting cooperation between 10GFC and
P802.3ae efforts. One of the proposals I will be making to P802.3ae is
the reservation of certain codes for 10GFC usage. This applies to the
10GFC Signal and Sequence Ordered-Sets. 

Best Regards,
Rich
     
--

"Bharadwaj S. Amrutur" wrote:
> 
> Rich,
> 
> Pulse ordered sets - is semantically extremely confusing.
> I would prefer the simpler  LS : Link status
> 
> You mention a desire to keep the running disparity of the D octets
> in the K/D/D/D  0. If we make sure that the LS code is inserted
> prior to  XAUI encoding and deleted after XAUI decoding,
> running disparity shouldn't be a problem, right?
> 
> You also mention a new ordered set K28.6/D/D/D and its translation into
> 0x0101 for 64/66. Was this proposed in the November plenary?
> 
> regards
> Birdy
                                 
------------------------------------------------------- 
Richard Taborek Sr.                 Phone: 408-845-6102       
Chief Technology Officer             Cell: 408-832-3957
nSerial Corporation                   Fax: 408-845-6114
2500-5 Augustine Dr.        mailto:rtaborek@xxxxxxxxxxx
Santa Clara, CA 95054            http://www.nSerial.com