Electrical computers and digital processing systems: multicomput – Computer network managing – Computer network monitoring
Reexamination Certificate
1998-04-10
2001-08-21
Sheikh, Ayaz (Department: 2155)
Electrical computers and digital processing systems: multicomput
Computer network managing
Computer network monitoring
C709S228000, C370S412000
Reexamination Certificate
active
06279035
ABSTRACT:
COPYRIGHT NOTICE
Contained herein is material that is subject to copyright protection. The copyright owner has no objection to the facsimile reproduction of the patent disclosure by any person as it appears in the Patent and Trademark Office patent files or records, but otherwise reserves all rights to the copyright whatsoever.
BACKGROUND OF THE INVENTION
1. Field of the Invention
The invention relates generally to the field of computer networking. More particularly, the invention relates to a flexible mechanism for reducing the amount of control plane processing and flow detection required in a Multi-protocol Over ATM (MPOA) system.
2. Description of the Related Art
With the emergence and growing popularity of Internet, intranet, client/server and multimedia applications, an increasing percentage of network traffic is now traversing subnet boundaries. Additionally, today's networks typically include a number of Local Area Networks (LANs) implementing diverse protocols. In response to these trends, the Asynchronous Transfer Mode (ATM) Forum Technical Committee has published the Multi-Protocol Over ATM (MPOA) specification, Multi-Protocol Over ATM Version 1.0, AF-MPOA-0087.000, published July 1997 (hereinafter “MPOA,” “the MPOA protocol,” or “the MPOA specification”). A network viewed at the ATM layer, allows any switch with an ATM interface to directly establish a circuit or connection to any other switch connected to the same ATM network. Higher level protocol traffic is typically constrained to flow through a router when crossing subnet boundaries. MPOA removes this constraint for the most part by efficient use of ATM circuits. MPOA facilitates the transfer of inter-subnet data by providing a framework in which internetwork layer protocols and other mechanisms for communicating across subnet boundaries, such as Internet Protocol (IP), IPv6, Internetwork Packet Exchange (IPX), DECnet routing, CLNP, AppleTalk, DDP, Vines, SNA, etc., may be efficiently overlaid on top of ATM.
Generally, the MPOA protocol provides a mechanism that greatly increases the efficiency of steady stream transmissions across subnet boundaries by identifying “flows” (e.g., a uni-directional flow of data packets to a single destination internetwork layer address) and mapping them onto ATM virtual channels. After detecting a flow, the MPOA protocol establishes a path called a “shortcut,” an ATM virtual channel connection (VCC), upon which data packets associated with the flow may be forwarded to avoid the hop-by-hop processing typically performed by intermediate routers along the “default path.”
A simplified MPOA system
100
is illustrated by FIG.
1
. The MPOA system
100
includes edge devices
110
and
140
, e.g., network devices, such as LAN-to-ATM switches or other MPOA devices, that are directly connected to an ATM cloud
150
and LAN hosts and/or LAN segments. The MPOA system
100
of the present example also includes one or more intermediate network devices, such as routers
120
and
130
, between edge devices
110
and
140
. It is appreciated that additional intermediate network devices, such as ATM switches and routers, may be located on the data path between router
120
and router
130
.
Edge devices
110
and
140
include LAN emulation clients (LECs)
114
and
144
, respectively and MPOA clients (MPCs)
112
and
142
, respectively. LECs perform forwarding in accordance with the ATM Forum's LAN Emulation Over ATM specification. Typically, network devices have a LEC for each emulated LAN (ELAN) interface. MPCs are MPOA protocol entities that implement the client side of the MPOA protocol. MPCs typically perform such functions as flow threshold detection, shortcut resolution, cache imposition request processing and handling of packets that arrive via a shortcut, each of which are described further below.
Routers
120
and
130
each include MPOA protocol entities that implement the server side of the MPOA protocol, MPOA server (MPS)
122
and
132
, respectively. For example, the MPSs are responsible for maintaining and distributing knowledge of the topology of the network. Additionally, the routers
120
and
130
include LECs
124
,
126
,
134
,
136
for the ELAN interfaces.
For purposes of this example, edge device
110
is assumed to be the point at which a stream of data enters the MPOA system
100
and edge device
140
is the point at which the flow exits the MPOA system
100
. For example, end-station
116
coupled to a LAN port (not shown) of edge device
110
may be transmitting packets to end-station
146
coupled to a LAN port (not shown) of edge device
140
. Further, this example assumes that end-stations
116
and
146
are on different subnets. Therefore, MPC
112
operates in its role as an Ingress MPC (I-MPC), MPS
122
operates in its role as an Ingress MPS (I-MPS), MPS
132
operates in its role as an Egress MPS (E-MPS), and MPC
142
operates in its role as an Egress MPC (E-MPC).
An Ingress Cache (I-Cache)
170
is maintained by MPC
112
for purposes of detecting inbound flows and keeping track of the shortcut VCC and encapsulation information (e.g., the LLC header to prepend to a packet before sending it on the shortcut) for those flows. Typically, an inbound flow is determined to exist once the MPC counts a predetermined number of packets addressed to a specific end-station within a predetermined time interval. In this example, the MPC
112
performs flow threshold detection by creating I-Cache entries and keeping packet counts for each MPS/IP address pair.
An Egress Cache (E-Cache)
180
is maintained by MPC
142
to facilitate handling of packets received on shortcuts that are to be forwarded on an outbound LAN port. E-Cache entries include, among other things, encapsulation information (e.g., the outbound DLL header to prepend to the packet before sending it to the outbound port). Encapsulation information is entered into the E-Cache
180
at the direction of the E-MPS
132
by way of a Cache Imposition Request protocol data unit (PDU)
182
.
While for purposes of explanation, MPOA protocol entities are depicted as residing on separate devices, it is appreciated that two or more MPOA protocol entities may be co-located. A LAN-LAN flow, for example, may involve two MPCs that reside on the same edge device, one serving as the Ingress MPC and the other as the Egress MPC. Additionally, a single MPOA protocol entity may assume the role of both an ingress and an egress for a particular transmission path. For instance, edge devices
110
and
140
might be separated by only a single router
120
or
130
, in which case, the MPS
122
or
132
would perform both ingress and egress MPS functions for data packets sourced at end-station
116
for end-station
146
. Further, it is important to note that MPOA protocol entities are logical rather than physical entities and therefore may span one or more physical devices.
Exemplary MPOA protocol scenarios and associated data and control plane processing will now be briefly described. Initially, the I-Cache
170
and E-Cache
180
have no entries. Upon receiving a first packet at edge device
110
that is destined for end-station
146
(i.e., a packet containing the internetwork address of end-station
146
and the MAC address of router
120
), an entry
171
is created in the I-Cache
170
and a packet count for this path is initialized to 1. The first packet destined for end-station
146
is forwarded via normal LANE procedures through LEC
114
over the default path (i.e., the hop-by-hop path from router
120
to router
130
as determined by routing protocols). Subsequent data packets destined for end-station
146
cause the packet count associated with the MAC/internetwork address pair to be incremented. That is, each packet being sent to an MPS is tallied by its destination internetwork address. These subsequent packets continue to be forwarded over the default path until an inbound flow is detected. A “flow” is said to be detected when the number of packets to a destination within a predetermined time interva
Brown Brian
Haney Jeanne
Mangin James
Pitcher Derek H.
Seshadri Kishore K.
Blakely , Sokoloff, Taylor & Zafman LLP
Dinh Khanh Quong
Nortel Networks Limited
Sheikh Ayaz
LandOfFree
Optimizing flow detection and reducing control plane... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Optimizing flow detection and reducing control plane..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Optimizing flow detection and reducing control plane... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2462332