Programmable reassembly of data received in an ATM network

Details Programmable reassembly of data received in an ATM network Programmable reassembly of data received in an ATM network

2000-05-09

2001-12-04

Olms, Douglas (Department: 2661)

Multiplex communications

Communication techniques for information carried in plural...

Assembly or disassembly of messages having address headers

C370S395430

Reexamination Certificate

active

06327271

ABSTRACT:

FIELD OF THE INVENTION
The present invention relates generally to data communications and the interface between a computer system and a network, and more particularly to methods and apparatus for efficiently receiving an asynchronous transfer mode (ATM) protocol data unit (PDU) over an ATM network.
BACKGROUND OF THE INVENTION
Asynchronous transfer mode (ATM) configured networks allow high-speed data, voice and video communications to be conducted between endpoint computer systems. ATM networks, which are based on the transmission of fixed-length data packets, have proven to be extremely useful because they combine the benefits of both a switched network (i.e., constant transmission delay, guaranteed capacity) and a packet switched network (i.e., flexibility and efficiency for intermittent traffic).
Current ATM standards are defined by the International Telecommunication Union and (ITU) ATM Forum specifications, which are herein incorporated by reference. As is common in contemporary communications protocols, several protocol layers are used to functionally divide the communications task within an ATM network. The ATM protocol layers are similar in scope to the Open System Interconnection (OSI) reference model that is defined by the International Standardization Organization (ISO).
In ATM networks, a variable length protocol data unit (PDU) defines the data to be shared between higher protocol layers, such as the application layer software programs operating at the endpoint computer systems. A typical PDU includes the data to be shared along with additional header and trailer information. To transmit the PDU over an ATM configured network, each PDU is further divided into fixed-length transmission units, known as cells. A typical cell is 53 bytes long and includes a 5-byte header containing its' connection identifier and a 48-byte payload. Thus, a 480-byte PDU would be divided into ten cells, each cell having a 48 byte payload, or one tenth of the total PDU.
During transmission, a cell is sent from one endpoint computer system to another through a virtual circuit within the interconnecting ATM network. A virtual circuit acts as a logically independent connection with another network node. A virtual circuit typically is a concatenation of communication links established between the two endpoints where higher layer protocols are accessed. By definition, ATM cells are transmitted in a sequence over an established virtual circuit. As such, the virtual circuit exists throughout the transmission of a PDU. One of the advantages of an ATM configured network is that a number of virtual circuits can be established over a single wire or fiber connecting the sending computer system to the network by time-division multiplexing the cells from different PDUs.
Typically, an ATM Network Interface Card (NIC) and accompanying software are provided within the sending (or receiving) endpoint computer systems to transmit (or receive) the cells of a PDU over a virtual circuit. In terms of the OSI reference protocol model, a typical NIC provides link layer functionality by supplying cells in a specific sequence to the physical layer of the ATM network. In contrast, the virtual circuits within the ATM network are typically established at a higher level layer, as are the PDUs and information therein.
FIG. 1A
is a block diagram illustrating a typical ATM network
10
having a first endpoint computer labeled host
12
, a network
14
, and one or more additional endpoint computers labeled end stations
16
. Within network
14
there are illustrated, by way of dashed connecting lines, a plurality of virtual circuits
18
that represent the communication channels established between host
12
and end stations
16
during an ATM communication. By way of example, network
14
may include one or more telecommunications and/or data networks, having switching devices, routing devices, and dedicated communication lines and/or fibers that are capable of providing a communication link between host
12
and end stations
16
. Host
12
and end stations
16
may, for example, be personal computer systems, workstations, mainframes, or other like processing devices that are capable of sending and receiving ATM PDUs.
FIG. 1B
is a block diagram that illustrates one possible configuration of an endpoint computer system (such as host
12
in
FIG. 1A
) having a processor
20
, a host bus
22
, a system memory
24
, a PCI controller
26
, a PCI bus
28
, a NIC
30
, and an optional SCSI interface
32
and SCSI device
34
. Processor
20
may be a microprocessor or central processing unit (CPU) configured to access system memory
24
. System memory
24
may be a dynamic random access memory (DRAM) that is accessed via host bus
22
or by way of another interconnecting circuit. SCSI device
34
may be a secondary data storage device, such as a disk drive unit, that can be accessed by processor
20
by way of host bus
22
, PCI controller
26
, PCI bus
28
, and SCSI interface
32
. As shown, processor
20
can also access network
14
by way of PCI bus
28
and NIC
30
. It is recognized that additional processors, other devices, additional buses, etc., can be connected to either the host bus or to the PCI bus as is common in modern computing configurations.
In a typical endpoint computer system, when the higher level protocol and/or application layers require a PDU to b e transmitted over network
14
to another endpoint computer system several process steps occur. First, a virtual circuit is established by processor
20
via NIC
30
. Next, the PDU is stored in system memory
24
by processor
20
. Following that, NIC
30
is directed by processor
20
to complete the desired PDU transmission. To complete the transmission of the PDU, NIC
30
fetches the cells within the PDU and transmit these cells one-by-one over a virtual circuit in network
14
.
On the receiving end of th e ATM network is another endpoint computer. This endpoint computer also includes a NIC
30
that receives the incoming cells in the same order as they were transmitted. As the cells are received, it is the task of NIC
30
and/or processor
20
to reassemble the cells into th e original PDU. The reassembled PDU can then be provided to the higher layer applications. As such, one important consideration in the design and operation of a NIC and its associated software is the reassembly of the PDU and the transferring of either the received cells and/or a reassembled PDU from the NIC to the system memory in the computer. A NIC that is able to effect the transfer of a PDU to system memory efficiently is valuable.
Conventional NICs can b e divided into one of two groups based upon their reassembly methods. In the first reassembly method, the NIC provides each cell as it arrives to the system memory. The NIC manages reassembly buffers in system memory, sending a received cell to the appropriate buffer based on its virtual circuit identifier and thereby reassembling the PDU. This type of reassembly is used, for example, in the IDT77201 product available from Integrated Device Technology Inc. of Santa Clara, Calif. While this type of operation tends to reduce the amount of memory required on the NIC, it also tends to place an unacceptable load or “transfer burden” on the interconnecting devices and buses which are required during the sequential transferring of each of the received cells from the NIC to the system memory. That is, the overhead with transferring a cell at a time from the NIC to system memory may be high.
The second reassembly method attempts to r educe the burden on the interconnecting devices and buses by providing enough memory on-board the NIC to store a complete PDU before transfer. Thus, with this type of NIC the entire PDU is reassembled in the NIC's local memory and then is transferred at one time from the NIC to the host memory. This type of reassembly is used, for example, in the 5930 product available from Adaptec Inc. of Milpitas, Calif. While this type of NIC tends to reduce the “transfer burden” by minimizing the amount o

Affiliated with

Also associated with

Rating

Programmable reassembly of data received in an ATM network does not yet have a rating. At this time, there are no reviews or comments for this patent.

If you have personal experience with Programmable reassembly of data received in an ATM network, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Programmable reassembly of data received in an ATM network will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFUS-PAI-O-2569452