Electrical computers and digital data processing systems: input/ – Intrasystem connection – Bus interface architecture
Reexamination Certificate
2002-11-20
2004-01-13
Auve, Glenn A. (Department: 2181)
Electrical computers and digital data processing systems: input/
Intrasystem connection
Bus interface architecture
C710S317000, C370S412000
Reexamination Certificate
active
06678778
ABSTRACT:
FIELD OF THE INVENTION
This invention relates generally to computer networks. In particular it pertains to a relatively low cost network switching hub and bus structure integrated into an integrated circuit chip, which improves network performance by increasing bandwidth and reducing collisions.
Personal computers (PCs) have permeated nearly all levels of society and business, creating a need to link individual PCs into networks to more efficiently utilize and share resources. Computer networks are becoming increasingly popular in low cost, performance oriented computing environments.
One very popular type of network used to link PCs or workstations is called a local area network (LAN). LANs and other types of networks permit the sharing of resources such as software, printers, modems, and other peripherals among PC workstations operating as nodes on the network.
FIG. 1
illustrates a simple and conventional LAN, generally designated by the reference numeral
100
. Network
100
comprises a plurality of PC workstations
112
a
through
112
e
, each connected by a communication link
113
a
through
113
e
respectively to an associated hub
114
. Communication links in such networks are typically two conductor cables, wherein a potential across the two conductors is varied in a manner representing sequential binary data. Such a link is termed a serial link.
Hubs, such as hub
114
, in networks are used to connect multiple workstations for routing through a single link to a server. In
FIG. 1
, PC workstations
112
a
through
112
e
are all connected to server
116
through link
115
. Hubs typically have a limited number of input ports, so the number of workstations that can be connected is limited as well. Typically the input ports are arranged in groups of 8, 16, 32, 64 and so forth. Further network capacity can be added by connecting multiple hubs, such as hub
118
(shown in dashed lines) to the one file server, and the additional hubs may then be connected to other multiple workstations (not shown). Other network variations include addition of multiple file servers connected in different arrangements to multiple hubs.
File server
116
in this example, and generally in the art, comprises a set of sheared high capacity mass storage devices, such as hard disks. Such file servers are often special PCs that have higher performance capability and more and larger capacity hard disks than do individual workstations
112
a
through
112
e
. The shared disk space on such file servers typically stores software applications which spreads the cost of the hard disk over more than one user, thereby permitting more efficient use of resources.
File server
116
in this example may also contain routers (not shown) for communication and connection to different network protocols such as Ethernet™, Asynchronous Transfer Mode (ATM), and Fiber Distributed Data Interface (FDDI), among others. The output of file server
116
is coupled to shared peripherals such as a network modem
119
, a laser printer
120
, and other peripherals represented by element number
122
. All workstations
112
a
through
112
e
on the network share access to the peripherals connected to server
116
.
It will be apparent to those with skill in the art that the example of
FIG. 1
is but one of many network arrangements known in the art.
There are some limitations of a conventional LAN
100
as described above. For example, communication over link
115
is shared by all of the workstations, and if many workstations are attempting to communicate at once, bandwidth may be a problem, slowing communication. By way of example, coaxial lines used in many networks have a maximum data transfer rate of 10 megabytes per second (Mb/s). The maximum data transfer rate, which is related to bandwidth, ultimately determines the maximum number of workstations that can be adequately handled by the network. Overloading a network can result in lost connections, communication delays, slow system response, timeouts, and slow file transfer times. All of these situations decrease the efficiency of the network, and become very annoying to network users.
Another limitation of conventional LANs such as LAN
100
, is that communication collisions may occur between multiple workstations requesting access to the network. In commonly used network protocols such as Ethernet™, one way that collisions are handled is by processing one request at a time while buffering other requests in a first in first out (FIFO) buffer. Since access is granted one workstation at a time, other stations are required to wait, thereby decreasing efficiency. Networks with heavy traffic tend to have many collisions which may drop efficiency to unacceptable levels.
The problems of inefficient communication are exacerbated by addition of more workstations to the network and increased use of bandwidth hungry applications such as color publishing and document imaging. So networks that had adequate bandwidth when installed may be outdated simply by software development.
Another way to add bandwidth is by increasing the number of switching hubs. This solution often results in segmenting a single large network into multiple smaller networks, which decreases the amount of traffic that travels over any given communication link and thereby increases the bandwidth available to each individual user. Statistics and traffic patterns can be further analyzed for adjustments for optimal network performance.
Another disadvantage of conventional networks is the relatively high cost of multi port hubs. By way of example, a 32 or even 16 port hub for some networks can cost in the range of about 32K to 100K dollars, a substantial investment for any user. Also conventional hubs are typically separate units in an enclosure with a dedicated power supply and controlling electronics, adding to clutter and adding to cost.
What is needed is a switchable hub that is relatively low cost, compact, and increases network performance by increasing bandwidth and reducing collisions. As will be described hereinafter, the present invention provides a method and apparatus to meet these objectives.
SUMMARY OF THE INVENTION
In a preferred embodiment an integrated circuit (IC) switching hub is provided, comprising a parallel bus of n bus lines implemented as traces on the IC; a first data port adapter controller connected by n first port lines through a first output buffer to the bus, each first port line connected to a bus line source-to-drain through one of a set of first queue switch transistors, and connected to a first data port providing a first external data link for the switching hub; a first read amplifier connected by n first receiver lines one each directly to each one of the n bus lines, and by a first data link to the first port adapter controller; a second data port adapter controller connected by n second port lines through a second output buffer to the bus, each second port line connected to a bus line source-to-drain through one of a set of second queue switch transistors, and connected to a second data port providing a second external data link for the switching hub; a second read amplifier connected by n second receiver lines one each directly to each one of the n bus lines and by a second data link to the second port adapter controller; and an arbitrator controller connected by a first control line to the gates of the first set of queue switch transistors, by a second control line to the gates of the second set of queue switch transistors, by a third control line to the first port adapter controller, and by a fourth control line to the second port adapter controller. The arbitrator controller is adapted to switch data from the first and second port adapter controllers onto the bus through the first and second output buffers and the first and second sets of queue switch transistors by switching the gates of the transistors via the first and second control lines respectively, and adapted to transfer data from the bus to the first and second external data links via the first and second read ampli
Auve Glenn A.
Elonex IP Holdings LTD
Woodcock & Washburn LLP
LandOfFree
Integrated network switching hub and bus structure does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Integrated network switching hub and bus structure, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Integrated network switching hub and bus structure will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3253333