Multiplex communications – Diagnostic testing – Determination of communication parameters
Reexamination Certificate
1998-03-31
2001-10-02
Ngo, Ricky (Department: 2664)
Multiplex communications
Diagnostic testing
Determination of communication parameters
C709S224000, C709S226000, C370S255000, C370S400000
Reexamination Certificate
active
06298044
ABSTRACT:
BACKGROUND OF THE INVENTION
Network administrators are charged with the task of the design, support and expansion of a new or existing network topology. When problems with the network occur, network administrators must quickly and efficiently diagnose and eliminate the problem in order to reduce network downtime of the end users. Network performance can suffer for a variety of reasons, the most basic of which is an inefficient layout of the network topology. For example, as networks grow it is common for the network administrator to migrate end nodes that talk together to the same collision domain without thought of overall network optimization. As a result, it is easy for some collision domains to become overloaded with traffic while other collision domains in the network are under utilized.
Other reasons for a poor network design include: (1) a network being handed down from administrator to administrator, each administrator adding to the network on a piecemeal basis, without thought to the overall system design, (2) a poor initial base network design, (3) lack of proper equipment, and/or (4) lack of knowledge and training of the network administrator. Poor network design can have devastating effects on the productivity of the system users. Allowing the network to be expanded from a poor base design simply compounds the problem.
Today's network administrators have a plethora of technologies to help collect data illustrating the traffic patterns on the network. One tool used by network administrators to collect data is a LAN analyzer. Technologies including HP EASE and RMON may also be used to collect traffic pattern data. However, although current technologies provide useful information, they do not provide an analysis of the traffic pattern data that illustrate and recommend to the network administrator the best way to reorganize end nodes to achieve increased performance.
Although traffic data collected, for example by the LAN analyzer, is useful the amount of data output is so large, it often requires significant efforts by the network administrator to wade through, decipher and analyze. Because the industry trend is to have fewer network administrators survey more and more complex network topologies, network administrators will often not have an intimate understanding of the network topology. Further, a network administrator is often taking over a network previously supported by another network administrator and does not have the time to do complex network analysis. Current data collection applications provide no suggestion to network administrators as to how and when collision domains in the network topology should be split to optimize network performance. An optimization tool that interprets this data and suggests how the network should be changed to achieve an optimal network topology is needed.
SUMMARY OF THE INVENTION
The present invention provides an optimization tool that utilizes traffic data across collision domain boundaries to provide suggestions as to how the network should be changed to achieve an optimal network topology. The optimization tool identifies collision domains that are reaching their peak capacity and analyzes their traffic pattern data to determine whether these over utilized collision domain should be divided into smaller collision domains.
Because interconnecting devices such as bridges, routers, and switches may be store and forward devices that increase the latency of traffic passing through them, traffic crossing these devices should be kept to a minimum. As a result, the optimization tool ensures that in reducing utilization by dividing a collision domain into two or three smaller collision domains that other network problems are not introduced such as delays being caused by traffic spanning a switch. Also provided are other performance-improving suggestions for collision domains that cannot be subdivided.
Once traffic data is collected, the steps implemented by the optimization tool to determine the optimal network topology include: identifying and removing global talkers that are not inextricably tied to another node; creating workgroups based on traffic patterns; and inserting previously removed global talkers back into the analysis.
Global talkers are nodes whose traffic is dispersed among many different collision domains. The purpose of identifying the global talkers is that global talkers should be placed on a collision domain of their own on a faster speed media. Because placement of global talker nodes is already determined, global talkers are removed from the analysis before further network optimization analysis occurs. There is one exception to this rule, in the case where a global talker has a node inextricably tied to it, the global talker is not removed. In this case, placing the global talker on its own dedicated port is not efficient since the node tied to the global talker is so heavily dependent on it, that nearly all of its traffic will cross a bridge, switch, or router.
After removal of the global talkers, the next step in the optimization process is the creation of workgroups. Workgroups are logical groupings of nodes that talk to one another. Further, a workgroup is tied to a particular collision domain. To group the nodes into workgroups, dependencies need to be explicitly drawn.
One dependency considered is traffic flow between a node under consideration and the node that it talks to. A node is considered to have an owner if more than half its total traffic is directed to or from that owner. When creating workgroups, a node should be placed in the same workgroup as its owner.
Also when determining whether or not a collision domain can be divided into its workgroups, compare the amount of traffic on each workgroup to the amount of internal traffic on the collision domain. The reason for this determination is that if we decide to split this collision domain, we want the collision domain's traffic to be reasonably dispersed among the newly created collision domains. Otherwise, one of the new collision domains may get overloaded in a fairly short period of time.
If it has been determined that a collision domain cannot be split into new collision domains, either because only one workgroup has been identified on it, or because the two workgroups that were identified either didn't distribute the original traffic evenly or caused too much traffic between them, the next step is to consider alternatives to splitting a collision domain. There are two such alternatives: upgrade the collision domain's speed or move the whole collision domain to a desktop switch.
The optimization tool frees the network administrator from the tedious manual analysis of data. In essence, all analysis is implemented behind the scenes with a network-wide rather than a collision domain wide view and the final conclusions are laid out for the user.
A further understanding of the nature and advantages of the present invention may be realized with reference to the remaining portions of the specification and the attached drawings.
REFERENCES:
patent: 5450408 (1995-09-01), Phaal
patent: 5598532 (1997-01-01), Liron
patent: 5640384 (1997-06-01), Du
patent: 5712981 (1998-01-01), McKee et al.
patent: 5881048 (1999-03-01), Croslin
Duong Frank
Hewlett--Packard Company
Lee Denise A.
Ngo Ricky
LandOfFree
Method and apparatus for determining if overloaded collision... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method and apparatus for determining if overloaded collision..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and apparatus for determining if overloaded collision... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2599529