Multiplex communications – Network configuration determination – Using a particular learning algorithm or technique
Reexamination Certificate
1998-10-05
2001-12-04
Olms, Douglas (Department: 2661)
Multiplex communications
Network configuration determination
Using a particular learning algorithm or technique
C709S225000
Reexamination Certificate
active
06327252
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates to automatically establishing links between distributed data processing servers connected to a non-broadcast mutli-access (NBMA) communications network, such as an asynchronous transfer mode (ATM) network.
BACKGROUND OF THE INVENTION
In computer networks it is often advantageous to provide a distributed computer system consisting of multiple computers which work with each other in an automated and controlled manner, in order to improve system reliability or increase system performance, for instance, when compared to a single computer effecting similar functionality.
A key requirement for such distributed computing systems is the need to maintain communication links between the individual computers, which are commonly referred to as servers. Frequently this is accomplished by creating a mesh of connections whereby each server has a link to every other server in the group or cluster manifesting the distributed system, but this is often impractical due to the large number of links for even moderately sized clusters.
Known problems concerning the communications infrastructure to link the various servers include:
1) Discovery of servers;
2) Minimizing link cost;
3) Ensuring reliability;
4) Avoiding looping;
5) Minimizing link establishment time; and
6) Implementation and administrative complexity.
A technique that solves one or more of these problems is desirable.
SUMMARY OF THE INVENTION
It is an object of the present invention to provide a new and improved methodology for establishing links between multiple servers connected to a communications network.
The invention, therefore, according to a first exemplary aspect provides a method of automatically establishing a link between a newly initialized data processing server and distributed data processing servers connected to a communications network comprising: arranging the distributed data processing servers in an interconnected configuration; initiating a discovery addressing protocol by the newly initialized server to discover other servers within the interconnected configuration; and forming a communications link between the newly initialized computer and at least one of the distributed data processing servers in the interconnected configuration upon discovery thereof.
In a particular embodiment of this aspect of the invention, the discovery address is re-initialized for a set number of times in the event of a failure to discover a peer server. If a peer server is not discovered a hunt addressing protocol is initialized.
In accordance with a second exemplary aspect of the invention, there is provided a method of automatically re-establishing links between distributed data processing servers in a communications network in the event of network failure, the method comprising: arranging the servers in an interconnected configuration; initiating a hunt addressing protocol by servers in the interconnected configuration to locate peer servers; and establishing communication links between the located servers.
Distributed computing where the underlying communications infrastructure is an NBMA based communications network characterizes a preferred embodiment of the invention. Examples of such a network include an ATM network or other types of virtual connection oriented digital communications networks.
The automatic link establishment technique embodying the present invention results in the interconnection of a plurality of servers, for example in the form of a spanning tree, which advantageously is an arrangement having a minimum number of links and avoids loops. Advantages of the discovery process whereby new servers connect to the tree include minimum link cost at the time of connection (i.e., each server will be connected to its nearest peer), and administration of the process is relatively simple, in that there are only two required configuration parameters (anycast addresses), and the parameters are common to all servers for a group. The resulting tree is self-healing, in that the technique enables automatic rebuilding of the cluster after a link or server failure. No redundant links are required due to this self-healing capability, which is administratively simple to effect.
REFERENCES:
patent: 5574860 (1996-11-01), Perlman
patent: 5822320 (1998-10-01), Horikawa
patent: 5854901 (1998-12-01), Cole
patent: 6016319 (2000-01-01), Kshirsagar
patent: 6032194 (2000-02-01), Gai
patent: 6047329 (2000-04-01), Horikawa
patent: 6069895 (2000-05-01), Ayandeh
The ATM Forum Technical Committee; Private Network-Network Interface Specification Version 1.0 (PNNI 1.0); af-pnni-0055.000, Mar. 1996.
The ATM Forum Technical Committee; ATM User-Network Interface (UNI) Signaling Specification Version 4.0; af-sig-0061.000, Jul., 1996.
Bajpai Sirish
Silton Roderick P.
Alcatel Canada Inc.
Larson & Taylor PLC
Olms Douglas
Pizarro Richardo M.
LandOfFree
Automatic link establishment between distributed servers... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Automatic link establishment between distributed servers..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Automatic link establishment between distributed servers... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2581616