Network Working Group M.T. Rose
Internet-Draft Invisible Worlds, Inc.
Expires: January 11, 2001 July 13, 2000
Mapping the BXXP Framework onto TCP
draft-mrose-bxxp-tcpmapping-01
Status of this Memo
This document is an Internet-Draft and is in full conformance with
all provisions of Section 10 of RFC2026.
Internet-Drafts are working documents of the Internet Engineering
Task Force (IETF), its areas, and its working groups. Note that
other groups may also distribute working documents as
Internet-Drafts.
Internet-Drafts are draft documents valid for a maximum of six
months and may be updated, replaced, or obsoleted by other documents
at any time. It is inappropriate to use Internet-Drafts as reference
material or to cite them other than as "work in progress."
The list of current Internet-Drafts can be accessed at
http://www.ietf.org/ietf/1id-abstracts.txt.
The list of Internet-Draft Shadow Directories can be accessed at
http://www.ietf.org/shadow.html.
This Internet-Draft will expire on January 11, 2001.
Copyright Notice
Copyright (C) The Internet Society (2000). All Rights Reserved.
Abstract
This memo describes how a BXXP session is mapped onto a single TCP
connection.
Rose Expires January 11, 2001 [Page 1]
Internet-Draft Mapping the BXXP Framework onto TCP July 2000
Table of Contents
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 3
2. Session Management . . . . . . . . . . . . . . . . . . . . . 4
3. Data Exchange . . . . . . . . . . . . . . . . . . . . . . . 5
3.1 Flow Control . . . . . . . . . . . . . . . . . . . . . . . . 6
3.1.1 Channel Creation . . . . . . . . . . . . . . . . . . . . . . 6
3.1.2 Sending REQ or RSP Messages . . . . . . . . . . . . . . . . 7
3.1.3 Processing SEQ Frames . . . . . . . . . . . . . . . . . . . 8
3.1.4 Use of Flow Control . . . . . . . . . . . . . . . . . . . . 9
References . . . . . . . . . . . . . . . . . . . . . . . . . 10
Author's Address . . . . . . . . . . . . . . . . . . . . . . 10
A. Changes from draft-mrose-bxxp-tcpmapping-00 . . . . . . . . 11
B. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . 12
Full Copyright Statement . . . . . . . . . . . . . . . . . . 13
Rose Expires January 11, 2001 [Page 2]
Internet-Draft Mapping the BXXP Framework onto TCP July 2000
1. Introduction
This memo describes how a BXXP[1] session is mapped onto a single
TCP[2] connection. Refer to Section 2.5 of [1] for an explanation of
the mapping requirements.
Rose Expires January 11, 2001 [Page 3]
Internet-Draft Mapping the BXXP Framework onto TCP July 2000
2. Session Management
The mapping of BXXP session management onto the TCP service is
straight-forward.
A BXXP session is established when a TCP connection is established
between two BXXP peers:
o the BXXP peer that issues a passive OPEN call is termed the
listener; and,
o the BXXP peer that issues an active OPEN call is termed the
initiator.
A BXXP session is released when either peer issues the CLOSE call,
and the TCP connection is subsequently closed.
A BXXP session is terminated when either peer issues the ABORT call,
and the TCP connection is subsequently aborted.
Rose Expires January 11, 2001 [Page 4]
Internet-Draft Mapping the BXXP Framework onto TCP July 2000
3. Data Exchange
The mapping of BXXP data exchange onto the TCP service is less
straight-forward.
Messages are reliably sent and received using the SEND and RECEIVE
calls. (This also provides ordered delivery of messages on the same
channel.)
Although TCP imposes flow control on a per-connection basis, if
multiple channels are simultaneously in use on a BXXP session, BXXP
must provide a mechanism to avoid starvation and deadlock. To
achieve this, BXXP re-introduces mechanisms used by the TCP:
sequence numbers and window-based flow control -- each channel has a
sliding window that indicates the number of payload octets that a
peer may transmit before receiving further permission.
Rose Expires January 11, 2001 [Page 5]
Internet-Draft Mapping the BXXP Framework onto TCP July 2000
3.1 Flow Control
Recall from Section 2.2.1.2 of [1] that every payload octet sent in
each direction on a channel has an associated sequence number.
Numbering of payload octets within a frame is such that the first
payload octet is the lowest numbered, and the following payload
octets are numbered consecutively.
The actual sequence number space is finite, though very large,
ranging from 0..4294967295 (2**32 - 1). Since the space is finite,
all arithmetic dealing with sequence numbers is performed modulo
2**32. This unsigned arithmetic preserves the relationship of
sequence numbers as they cycle from 2**32 - 1 to 0 again.
3.1.1 Channel Creation
When a channel is created, the sequence number associated with the
first payload octet of the first frame is 0, and the initial window
size for that channel is 4096 octets. After channel creation, a BXXP
peer may update the window size by sending a "SEQ" frame (Section
3.1.3).
If a BXXP peer is requested to create a channel and it is unable to
allocate at least 4096 octets for that channel, it must decline
creation of the channel, as specified in Section 2.3.1.3 of [1].
Similarly, during establishment of the BXXP session, if the BXXP
peer acting in the listening role is unable to allocate at least
4096 octets for channel 0, then it must return a negative response,
as specified in Section 2.4 of [1]. instead of a greeting.
Rose Expires January 11, 2001 [Page 6]
Internet-Draft Mapping the BXXP Framework onto TCP July 2000
3.1.2 Sending REQ or RSP Messages
Before a message is sent, the sending BXXP peer must ensure that the
size of the payload is within the window advertised by the receiving
BXXP peer. If not, it has three choices:
o if the window would allow for at least one payload octet to be
sent, the BXXP peer may segment the message and start by sending
a smaller frame (up to the size of the remaining window);
o the BXXP peer may delay sending the message until the window
becomes larger; or,
o the BXXP peer may signal to its application that it is unable to
send the message, allowing the application to try again at a
later time (or perhaps signaling its application when a larger
window is available.)
The choice is implementation-dependent, although it is recommended
that the application using BXXP be given a mechanism for influencing
the decision.
Rose Expires January 11, 2001 [Page 7]
Internet-Draft Mapping the BXXP Framework onto TCP July 2000
3.1.3 Processing SEQ Frames
As an application accepts responsibility for incoming frames, its
BXXP peer should send "SEQ" frames to advertise a new window.
The ABNF for a "SEQ" frame is:
seq = "SEQ" SP channel SP ackno SP window CR LF
ackno = seqno
window = size
; channel, seqno, and size are defined in Section 2.2.1 of [1].
The "SEQ" frame has three parameters:
o a channel number;
o an acknowledgement number, that indicates the value of the next
sequence number that the sender is expecting to receive on this
channel; and,
o a window size, that indicates the number of payload octets
beginning with the one indicated by the acknowledgement number
that the sender is expecting to receive on this channel.
A single space character (decimal code 32, " ") separates each
component. The "SEQ" frame is terminated with a CRLF pair.
When a "SEQ" frame is received, if any of the channel number,
acknowledgement number, or window size cannot be determined or is
invalid, then the BXXP session is terminated without generating a
response, and it is recommended that a diagnostic entry be logged.
Rose Expires January 11, 2001 [Page 8]
Internet-Draft Mapping the BXXP Framework onto TCP July 2000
3.1.4 Use of Flow Control
The key to successful use of flow control within BXXP is to balance
performance and fairness:
o large messages should be segmented into multiple frames (e.g.,
the ideal BXXP segment size should be no larger than TCP's
negotiated maximum segment size minus some small constant);
o frames for different channels with traffic ready to send should
be sent in a round-robin fashion; and,
o each time a "REQ" or "RSP" message is received, a "SEQ" frame
should be sent whenever the window size is at least one half of
the available buffer space (if the transport service presents
multiple messages to a BXXP peer simultaneously, then a single
consolidating "SEQ" frame may be sent).
In order to avoid pathological interactions with the transport
service, it is important that a BXXP peer advertise windows based on
available buffer space, to allow data to be read from the transport
service as soon as available. Further, "SEQ" frames for a channel
should have higher priority than "REQ" or "RSP" messages for that
channel.
Finally, implementations may wish to provide queue management
facilities to the application using BXXP, e.g., channel priorities,
(relative) buffer allocations, and so on. In particular,
implementations should not allow a given channel to monopolize the
underlying transport window (e.g., slow readers should get small
windows).
Rose Expires January 11, 2001 [Page 9]
Internet-Draft Mapping the BXXP Framework onto TCP July 2000
References
[1] Rose, M.T., "The Blocks eXtensible eXchange Protocol
Framework", draft-mrose-bxxp-framework-01 (work in progress),
July 2000.
[2] Postel, J., "Transmission Control Protocol", RFC 793, STD 7,
Sep 1981.
Author's Address
Marshall T. Rose
Invisible Worlds, Inc.
1179 North McDowell Boulevard
Petaluma, CA 94954-6559
US
Phone: +1 707 789 3700
EMail: mrose@invisible.net
URI: http://invisible.net/
Rose Expires January 11, 2001 [Page 10]
Internet-Draft Mapping the BXXP Framework onto TCP July 2000
Appendix A. Changes from draft-mrose-bxxp-tcpmapping-00
o The IPR notice is changed to be in full compliance of Section 10
of RFC 2026.
o SEQ messages are now (correcty) called SEQ frames.
o In Section 3.1.4, the explanation of when to send a SEQ frame is
clarified.
o In Section 3.1.4, the illustration of queue management facilities
is expanded.
Rose Expires January 11, 2001 [Page 11]
Internet-Draft Mapping the BXXP Framework onto TCP July 2000
Appendix B. Acknowledgements
Dave Crocker provided helpful suggestions on the nature of flow
control in the mapping.
Rose Expires January 11, 2001 [Page 12]
Internet-Draft Mapping the BXXP Framework onto TCP July 2000
Full Copyright Statement
Copyright (C) The Internet Society (2000). All Rights Reserved.
This document and translations of it may be copied and furnished to
others, and derivative works that comment on or otherwise explain it
or assist in its implementation may be prepared, copied, published
and distributed, in whole or in part, without restriction of any
kind, provided that the above copyright notice and this paragraph
are included on all such copies and derivative works. However, this
document itself may not be modified in any way, such as by removing
the copyright notice or references to the Internet Society or other
Internet organizations, except as needed for the purpose of
developing Internet standards in which case the procedures for
copyrights defined in the Internet Standards process must be
followed, or as required to translate it into languages other than
English.
The limited permissions granted above are perpetual and will not be
revoked by the Internet Society or its successors or assigns.
This document and the information contained herein is provided on an
"AS IS" basis and THE INTERNET SOCIETY AND THE INTERNET ENGINEERING
TASK FORCE DISCLAIMS ALL WARRANTIES, EXPRESS OR IMPLIED, INCLUDING
BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF THE INFORMATION
HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED WARRANTIES OF
MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE.
Acknowledgement
Funding for the RFC editor function is currently provided by the
Internet Society.
Rose Expires January 11, 2001 [Page 13]