US20030167418A1 - Fault-tolerant server for collaborative computing - Google Patents

Abstract

A distributed collaborative computer system is provided that comprises a plurality of server computers interconnected via a high-speed link. Client computers can connect to any available server computer and start or join a conference hosted on either the server computer to which the client computer is connected or any other server in the system. As a result, the system and method of the present invention is easily scalable to support an arbitrary number of participants to a conference by merely adding the appropriate number of server computers to the system. In addition, by replicating the conference information on more than one server computer, the single point of failure limitation is eliminated. In fact, if a server hosting or participating in a conference malfunctions, the failure is detected by other server computers and the client computer is able to reconnect to the conference through a new server computer.

Description

CROSS-REFERENCE TO CD-ROM APPENDIX
• An Appendix containing a computer program listing is submitted on a compact disk, which is herein incorporated by reference in its entirety. The total number of compact discs including duplicates is two. Appendix A, which is part of the present specification, contains a list of the files contained on the compact disk. These listings contain material which is subject to copyright protection. The copyright owner has no objection to the facsimile reproduction by anyone of the patent document or the patent disclosure, as it appears in the patent and trademark office patent file or records, but otherwise reserves all copyright rights whatsoever.[0001]
BACKGROUND OF THE INVENTION
• 1. Field of the Invention[0002]
• The present invention relates generally to computer networks and, more particularly, to collaborative computing over a computer network.[0003]
• 2. Description of the Related Art[0004]
• Traditional collaborative computing tools allow computer users at different locations to communicate via a computer network and share documents or applications stored and/or executed on one the user's computers. While both peer-to-peer and client-server communication models have been used in the past, web-based collaborative tools generally employ a client-server model.[0005]
• For example, client-server application sharing (also discussed in the context of “distributed computing”) is described in U.S. Pat. No. 5,434,852 “Distributed Processing Architecture for Control of Broadband and Narrowband Communication Networks;” U.S. Pat. No. 5,887,170 “System for Classifying and Sending Selective Requests...;” and U.S. Pat. No. 6,038,593 “Remote Application Control for Low Bandwidth Application Sharing,” all incorporated herein by reference in their entireties. Other group communication techniques are described by Ulrick Hall and Franz J. Hauck, “Promondia: A Java-Based Framework for Real-time Group Communication in the Web,” Proceedings of Sixth International World Wide Web Conference (Apr. 7-11, 1997); Lane Boyd, “Taking Collaboration Into Orbit,” Computer Graphics World, Vol. 21, No. 9, p. 36 (September 1998); and Eric Ly, “Distributed Java Applets for Project Management on the Web,” IEEE Internet Computing Online, Vol.. 1, No. 3 (May/June 1997), all incorporated herein by reference in their entireties.[0006]
• International Telecommunications Union (ITU) Standard T.120 is a family of open standards that provides both communications and applications protocols to support real-time multipoint data communications for collaboration and conferencing, among other uses. This standard is outlined in A Primer on the T.120 Series Standard by DataBeam Corp. (May 14, 1997), incorporated herein by reference in its entirety.[0007]
• FIG. 1A is a block diagram illustrating the communication scheme used for an exemplary traditional[0008]collaborative computer system100. In FIG. 1A, client computers110n(where n=A, B, C . . . ) can connect to server computers120nover a global-area computer network130 (e.g., the Internet). As used herein, the numeral n appended to a reference number does not imply any correspondence among elements having different numerals (e.g.,client computer110A bears no relationship toserver computer120A). FIG. 1B is a block diagram illustrating the actual communications channels established between client computers110nand server computers120nto set up two conferences between users ofclient computers110A and110B on the one end and110C and110D on the other. As is readily apparent from inspection of FIG. 1B, each conference is handled by a single server computer120n.This model performs satisfactorily for conferences having a small number of participants and conferences that do not require fault tolerance. However, as the number of participants in a conference increases, the computing power of server computer120nbecomes a bottleneck. Furthermore, if the particular server computer120nthat is handling a conference malfunctions, the entire conference is disrupted (i.e., server computer120nrepresents a single point of failure for the entire system handling that conference). Accordingly, there is a need for an improved collaborative computing system.
BRIEF SUMMARY OF THE INVENTION
• The system and method of the present invention provide a distributed collaborative computer system that is scalable to handle an arbitrary number of conference participants and eliminates the server as the single point of failure in the system. This is accomplished by providing a plurality of server computers interconnected via one or more high-speed links. Client computers can connect to any available server computer and start or join a conference hosted on either the server computer to which the client computer is connected or any other server in the system. As a result, the system and method of the present invention is easily scalable to support an arbitrary number of participants to a conference by merely adding the appropriate number of server computers to the system. In addition, by replicating the conference information on more than one server computer, the system and method of the present invention eliminate the single point of failure limitation of prior art systems. In fact, if a server hosting or participating in a conference malfunctions, the failure is detected by other server computers and the client computer is able to reconnect to the conference through a new server computer.[0009]
BRIEF DESCRIPTION OF THE DRAWINGS
• The present disclosure may be better understood and its numerous features and advantages made apparent to those skilled in the art by referencing the accompanying drawings.[0010]
• FIG. 1A is a block diagram of a prior art collaborative computer system.[0011]
• FIG. 1B is a block diagram illustrating the connections established between the client and server computer of FIG. 1A during two conferences.[0012]
• FIG. 2A is a block diagram of a distributed collaborative computer systems, in accordance with some embodiments of the invention.[0013]
• FIG. 2B is a block diagram illustrating the connections established between the client and server computers of FIG. 2A during a conference.[0014]
• FIG. 3 is a block diagram of the software components of a distributed collaborative computer system, in accordance with some embodiments of the invention.[0015]
• FIGS. 4A, 4B, and[0016]4C are flow diagrams illustrating a start/join conference operation on the distributed collaborative computer system of FIG. 3.
• FIG. 5 is a flow diagram of the operation of the log server of FIG. 3.[0017]
• FIG. 6 is a flow diagram of the operation of the license server of FIG. 3.[0018]
• FIG. 7 is a flow diagram of the operation of an App server of FIG. 3.[0019]
• FIGS. 8, 9,[0020]10, and11 are flow diagrams illustrating the operation of the meeting manager of FIG. 3.
• FIG. 12 is a block diagram illustrating the software components of the client and server computers of FIGS. 2A and 2B.[0021]
• FIGS. 13A, 13B, and[0022]13C are flow diagrams illustrating the operation of the CB server and App servers of FIG. 3.
• FIG. 14 is a block diagram illustrating the communication channels established between two client computers of FIG. 3 during an on-line conference, in accordance with an embodiment of the invention.[0023]
• FIG. 15 is a flow diagram of an operation for transmitting data between the client computers of FIG. 14.[0024]
• FIGS. 16A and 16B are flow diagram illustrating a skip page operation used to control transmission of pages between a presenter's client computer and other participants' client computers, in accordance with some embodiments of the invention.[0025]
• FIG. 17 is a flow diagram of a client browser operation, in accordance with some embodiment of the invention.[0026]
• FIGS. 18A, 18B,[0027]18C1-3,18D1-2,19A,19B,20A,20B and20C are views of web pages displayed by client browser of FIG. 3 during operation of the distributed collaborative computer system of FIG. 3.
• The use of the same reference symbols in different drawings indicates similar or identical items.[0028]
DETAILED DESCRIPTION OF THE INVENTION
• FIG. 2A illustrates a distributed[0029]collaborative computing system200, in accordance to some embodiments of the invention.Client computers210n(where n=A, B, C . . . ) are connected toserver computers220nvia global-area computer network130. Unlike in the prior art system of FIGS. 1A and 1B, eachclient computer210ncan connect to anyserver computer220n.Server computers220nare in turn connected through a high-speed link230.High speed link230 allows faster throughput and a higher level of security than global-area network130. For example, in some embodiments high-speed link130 is a dedicated T1 or T3 or optical carrier-class link, such as one employing the well-known SONET standard and OC-48 or OC-192 framing. One of ordinary skill in the art will readily recognize that many other equivalent high-speed network standards, including non-optical standards, may be employed to create a high bandwidth link.
• FIG. 2B illustrates the connections established between[0030]client computers210nandserver computers220nto conduct a conference between participants seated atclient computers210A and210D, respectively. First,client computer210A (whose user will host the conference) establishes aconnection225A toserver computer220A over global-area network130.Server computer220A, in turn, is connected toserver computer220B via high-speed link230. Finally,client computer210D, whose user will join the conference hosted by the user ofclient computer210A, establishes aconnection225B toserver computer220B over global-area network130. As a result, information transmitted fromclient computer210A travels throughconnection225A, high-speed link230 andconnection225B to reachclient computer210D. Similarly, information transmitted fromclient computer210D travels throughconnection225B, high-speed link230 andconnection225A to reachclient computer210A. Since high-speed link230 is several orders of magnitude faster thanconnections225A and225B, the delay introduced by high-speed link230 is transparent to the users ofclient computers210A and210B.
• FIG. 3 is a block diagram of the software components of a distributed[0031]collaborative computer system300, in accordance with some embodiments of the invention.
• Distributed[0032]collaborative computer system300 includes meeting zones310n(where n=A, B, C),client browser320,web zone330 andcentral operation database350.Client browser320 is a web browser program executed on one ofclient computers210n(FIGS. 2A and 2B).Client browser320 first connects toweb zone330 to request starting or joining a conference.Web zone330, in turn, verifies the user and conference information and updates central operation database340 accordingly. Onceweb zone330 has verified that the user is authorized to start/join a conference,client browser320 connects to one of meeting zones310nto access the conference. Meeting zone310n,in turn, connectsclient browser320 to the desired conference and updates central operation database340 accordingly.
• [0033]Web zone330 includes aweb server335 that maintains a website to allow users to access distributedcollaborative computer system300 and aweb database337 that stores web usage and administrative information about users of distributedcollaborative computer system300. The information stored inweb database337 is periodically synchronized and/or replicated with the information stored in central operation database340 to ensure data consistency.
• Each meeting zone[0034]310n,in turn, includes ameeting manager350, aping server355, alicense manager360, ameeting database365, alog server370, collaboration (CB) servers380n,and application (App)servers390n.Furthermore, each meeting zone310nalso includes a process manager (PM)311.Process manager311 is the controlling process for all logical servers running on a physical server within the meeting zone.PM311 thus monitors the health of all logical servers and processes running on the physical server and spawns replacement processes on failure. Alternatively,PM311 can start new processes on command from remote access service (RAS)312.
• In one embodiment of the present invention, a single instance of meeting[0035]zone310A is implemented on one physical server (i.e., one machine).
• In some embodiments, each meeting zone is implemented on a single physical server. One of ordinary skill will readily appreciate, however, that multiple physical servers could also be used either as hot or warm standby units for redundancy or to spread the logical server loading across multiple machines, each with its own PM. Alternatively, several meeting zones could be implemented on one physical server, either having their own PM, or sharing a single PM.[0036]
• [0037]PM311 spawns each logical server (e.g.,CB servers380A,380B,380C;App server390A,390B,390C;meeting manager350, ping server35,log server370, and license manager360) as directed by a startup configuration file or operator command throughRAS312.RAS312 is, in some embodiments, a real-time messaging service such as TIBCO Rendezvous, available from TIBCO Software, Inc. of Palo Alto, Calif.
• Each logical server has its own communications and control module known as a zone manager (ZM). Conceptually, each[0038]ZM313 is functionally similar although one of ordinary skill in the art will appreciate that implementation optimizations may allow for reduced functionality in some instances ofZM313.
• [0039]Meeting manager350 also possesses aspecial zone manager314, so designated because it also acts as a gatekeeper (GK) for the entity meeting zone310. The GK maintains a subset of the state of each logical server so that meetingmanager350 has immediately available the detailed status of the entire meeting zone310.
• Each ZM, which is spawned (created) in direct correspondence to each logical server or autonomous process on a given physical server machine, monitors the health and status of its corresponding logical server or process. All logical server communications with other logical servers and with the process manager[0040]211 go through the ZM in each logical server and the PM.
• The operational functions of[0041]PM311,RAS312,ZM313, and ZM/GK314 are discussed in further detail below.
• All ZMs report to a single “super ZM”, known as the gatekeeper or ZM/GK. Each ZM sends a subset of its corresponding logical server's state and traffic capacity to the ZM/GK so that the GK is aware of the status of all elements of the meeting zone. This enables the meeting manager to get coordinated zone state reports and therefore “know” the status of the entire meeting zone.[0042]
• Zone status is important to the meeting manager (and thus to the overall health and efficiency of the zone) because the meeting manager uses ZM/GK state reports to manage both the zone's overall quality of service (QoS) and the load balance across all active collaboration servers (CBs) in the zone.[0043]
• QoS, in this context, refers to the zone's ability to respond to client data requests of all types (e.g., HTTP, application sharing, document sharing, telephony, and so forth). In addition, QoS is an indirect indicator of latency to those requests, caused by high and possibly unbalanced loading of the logical servers in the meeting zone. For example, in some embodiments of the present invention, a meeting manager faced with a need to add more user participants to an in-progress meeting must determine if an additional CB server must be spawned (i.e., brought on-line) to keep overall CB server loading below a certain threshold. This “intelligence” in the MM is implemented through the ZMs in each CB and the coordinating function of the ZM/GK reporting to the MM. The MM can thus decide if the pre-defined QoS for the specific user client (perhaps determined by the time of day, the user's license, or the type of service purchased by the user or some communication thereof, to name but a few examples), would be unobtainable without additional CB server resources. If so, the meeting manager will request that the process manager spawn a new CB server.[0044]
• Once[0045]client browser320 has received authorization to start/join a conference,client browser320 attempts to connect to pingservers355 in multiple meeting zones310n.Client browser320 selects the first ping server to respond to the connection request and disconnects other respondingping servers355. The selected ping server, in turn, forwards the request to start/join a conference to ameeting manager350 in the same meeting zone310nas the selectedping server355.Meeting manager350, in turn, assigns a CB server380nto host/handle the conference. The selected CB server380nconnects toclient browser320 and any other CB servers380nhandling the conference that the user wishes to start/join. Thus,client browser320 communicates withother client browsers320 via the selected CB server380n.
• [0046]App servers390nare used by CB servers380nandclient browsers320 to support services such as document view, file sharing, video, voice over IP, telephony, polling, chat and application sharing. Collaborative support for these services are further described in the following references, each incorporated herein by reference in its entirety:
• “Instant Document Sharing,” co-pending and commonly-assigned Application for a U.S. patent Ser. No. 09/442,424, filed Nov. 17, 1999.[0047]
• “Instant Sharing of Documents in a Remote Server,” co-pending and commonly-assigned Application for U.S. patent Ser. No. 09/471,938, filed Dec. 23, 1999.[0048]
• “Remote Document Serving,” co-pending and commonly-assigned Application for a U.S. patent Ser. No. 09/591,377, filed Jun. 9, 2000.[0049]
• “Instantaneous Remote Control of an Unattended Server,” co-pending and commonly-assigned Application for a U.S. patent Ser. No. 09/515,684, files Feb. 29, 2000.[0050]
• “Method for Creating Peer-to-Peer Connections Over an Interconnected Network to Facilitate Conferencing Among Users,” co-pending and commonly-assigned Application for a U.S. patent Ser. No. 08/609,025, filed on Feb. 29, 1996.[0051]
• “Method for Establishing a Communication Connection Between Two or More Users Via a Network of Interconnected Computers,” co-pending and commonly-assigned Application for a U.S. patent Ser. No. 09/195,801, filed on May 12, 2000.[0052]
• “Emulating a Persistent Connection Using HTTP,” co-pending and commonly-assigned Application for a U.S. patent Ser. No. 09/449,011, filed on Nov. 24, 1999.[0053]
• “Method of Transferring Data at Adjustable Levels of Priorities to Provide Optimum Response to User Demands,” U.S. Pat. No. 5,623,603.[0054]
• “Method to Provide for Virtual Screen Overlay,” U.S. Pat. No. 5,577,188.[0055]
• “Collaborative Web Browser,” U.S. Pat. No. 5,944,791.[0056]
• [0057]Log server370 communicates with meetingmanager350 via theirrespective ZMs313 and314 and stores information related to new users joining/leaving conferences andupdates meeting database365.License manager360 communicates with meeting manager350 (again, throughZMs313 and314) andpolls meeting database360 to ensure that the number of users authorized to join a meeting is not exceed.
• Overall fault tolerance is ensured by providing process-level fault monitoring by the ZM and correction (e.g., process replacement) by the PM. At the logical server level, the MM uses ZM/GK sate monitoring to detect logical server faults and PM commands to spawn replacements. Logical server state replication is also provided by the gatekeeper, using the meeting database. Finally, physical server fault tolerance is provided by operator hardware and environmental status using a combination of manual and RAS monitoring and control methods well-known in the art.[0058]
• FIGS.[0059]4A-4C are flow diagrams illustrating a start/join conference operation400 on distributed collaborative computer system300 (FIG. 3).
• First, in[0060]stage402,client browser320 connects to aweb server335. If the connection is successful (stage404),operation400 proceeds to stage406, otherwise stages402 and404 are repeated. Instage406, the user ofclient computer320 logs on toweb server335. Instage408, the information entered by the user instage406 is authenticated with information stored inweb database337. If the information entered by the user cannot be authenticated, stages406 and408 may be repeated until the information entered by the user is successfully validated. In some embodiments,client browser320 is disconnected after a predetermined number of login attempts to prevent unauthorized access toweb server335. As those skilled in the art are familiar with techniques for preventing/deterring unauthorized access to a website, these techniques are not further discussed herein.
• Once the user has successfully logged on to[0061]web server335,stage410 determines whether the user is requesting to start a new conference or join an existing conference. If the user is requesting to join a new conference,operation400 proceeds to stage412, otherwiseoperation400 proceeds to stage450.
• In[0062]stage412, meeting parameters are extracted from meetingdatabase365 throughweb database337. Instage414, a plug-in forclient browser320 is launched onclient computer210n(FIGS. 2A and 2B). The first time the user ofclient browser320 connects toweb server335, the plug-in is downloaded over global-area network130 and installed on theclient computer210n. After the plug-in is installed onclient computer210n, it can be re-used for subsequent conferences without the need for downloading and reinstalling it. In some embodiments, multiple versions of the plug-in are used over time: when a new version of the plug-in becomes available onweb server335, the new plug-in is downloaded toclient computer210nand installed in place of the older version of the plug-in.
• In[0063]stage416, the meeting parameters are sent from meeting database365 (via web database337) toclient computer210nandoperation400 proceeds to stage418 (FIG. 4B).
• In[0064]stage418,client browser320 attempts to connect to anyavailable ping server355. Instage420, responses are received from one ormore ping servers355. In some embodiments, if no response is received within a predefined time limit, stages418 and420 are repeated until a response is received within either the original time limit or a newly defined time limit.Client browser320 selects thefastest ping server355 to respond to the connection request (stage422) and disconnects the non-selected ping servers355 (stage424).Client browser320 then sends a request to join a meeting to the selected ping server355 (stage426) andping server355 forwards the request to a meeting manager (MM)350 (stage428) in the same meeting zone310n(FIG. 3) asping server355.
• Upon receiving the request to join a meeting,[0065]meeting manager350 selects a collaboration (CB) server380nfrom a pool of available CB servers380nin the meeting zone310n(stage430). In stage432 (FIG. 4C), the selected CB server380nqueries other CB servers380nin all meeting zones310nto ascertain which CB server380nis hosting the meeting to which the user ofclient browser320 is attempting to connect. Once client CB server380nlocates the hosting CB server380n, it connects to the hosting CB server380n(stage434). Client CB server380nthen makes a local copy of the meeting data from hosting CB server380n.
• [0066]Stage438 determines whether meetingmanager350 has received a meeting confirmation from client CB server380n, in whichcase operation400 proceeds to stage440. Otherwise stages418-438 are repeated with a new client CB server380n.
• In[0067]stage440,meeting manager350 has received confirmation from CB server380nthat a connection has been successfully established with the hosting CB server380n. The confirmation is then transmitted from meetingmanager350 toping server355 and fromping server355 to client browser320 (stage442).
• If the user requests starting a new meeting in[0068]stage410,operation400 proceeds to stages450-472. Stages450-466 are analogous to stages414-430 and stages468-472 are analogous to stages438-442, except that ifstage468 fails,operation400 proceeds to stage454 rather thanstage418.
• FIG. 5 is a flow diagram of the[0069]operation500 oflog server370 of FIG. 3. Inoperation500,stage510 determines whether a new log entry has been posted andstage520 updates meeting database365 (FIG. 3).
• FIG. 6 is a flow diagram of the[0070]operation600 oflicense server360 of FIG. 3. First,stage610 determines if a new user has requested joining the meeting, in whichcase operation600 proceeds to stage620. Otherwise,stage610 is repeated. Instage620,license manager360 compares the number of users in the meeting if the current user were allowed to join the meeting to the user limit for the meeting.Stage630 then determines whether the user limit is exceed, in which case CB server380nis notified (stage640). Otherwise stages610-630 are repeated.
• FIG. 7 is a flow diagram of the[0071]operation700 of an application (App)server390nof FIG. 3. First,App server390nregisters with meetingmanager350 in the same meeting zone310n(FIG. 3) instage710.Meeting manager350, in turn, allocatesApp server390nto a CB server380nhandling a given conference (stage720). CB server380n, in turn, initializesApp server390nwith the necessary application data required for the conference (stage730) and establishes a connection toApp server390n(stage740) viaZMs313. CB server380nnotifiesApp server390nof meeting events (e.g., users joining/leaving the meeting or control passing from the host to another user) instage750. Finally,App server390nestablishes a connection withclient browser320 via CB server380n(stage760) which allows users ofclient browsers320 to access and interact with the application provided byApp server390n.
• FIGS.[0072]8-11 are flow diagrams illustrating the operation of meeting manager (MM)350 for providing fault tolerance to distributedcollaborative computer system300.
• FIG. 8 illustrates CB server failure detection and[0073]recovery operation800. First,meeting manager350 checks whether any CB servers380nin the meeting manager's meeting zone310nhave failed (stage810). A variety of techniques known in the art can be employed to detect failure of CB servers380n. For example, CB servers380ncan periodically transmit a “heartbeat” message to meetingmanager350. If meetingmanager350 does not receive a heartbeat message from a CB server380nwithin a predefined time limit,meeting manager350 attempts to contact CB server380nand if no response is received from CB server380nwithin a predefined time limit,meeting manager350 determines that CB server380nhas failed. Other failure detection techniques known in the art can be used to detect failure of a CB server380nin accordance one or more embodiments of the present invention. Accordingly, the present invention is not limited to any particular failure detection technique.
• In some embodiments of the present invention,[0074]meeting manager350 employs its zone manager (and meeting zone gatekeeper) (ZM/GK)214 to exchange heartbeat (or analogous) messages withZM313 in each CB server380n. When and if ZM/GK314 detects a CB server (or other logical server failure) by noting a lack of heartbeats, for example, ZM/GK sends a request to process manager (PM)311 to restart the dead logical server.
• [0075]PM311 also monitors eachZM313, including ZM/GK314, to evaluate ZM health. ShouldPM311 discover a failed or stopped ZM process, the PM will restart (i.e., spawn a replacement for) the ZM.
• In particular, if failure of a CB server[0076]380nis detected instage810,operation800 proceeds to stage820. Otherwise stage810 is repeated until a failure is detected.Meeting manager350, in turn, retrieves a list of meetings handled by failed CB server380nfrom meeting database365 (stage820) and sends a request to processmanager311 to launch a new CB server380n(stage830).
• The newly-spawned (replacement) CB server recovers its state information (e.g., information describing its configuration, operating or quality of service [QoS] parameters, and/or current meeting data) from the local meeting zone's gatekeeper. Typically, this is the ZM/[0077]GK314 withinzone manager350, but the gatekeeper function may alternately be provided by any designatedZM313. Generally speaking, all local state in a logical server is preserved. However, if an application server goes down, the application state is lost. Only the meeting state is preserved in this case.
• [0078]Stage840 then determines if the new CB server380nhas successfully come on-line, in whichcase meeting manager350 continues to monitor the status of CB servers380n(stage810). Otherwise, stages830-840 are repeated until a new CB server380nsuccessfully comes on-line.
• FIG. 9 illustrates the application server failure detection and[0079]recovery operation900. First,meeting manager350 and CB servers380n(FIG. 3) check whether anyApp servers390nin the same meeting zone310nas meetingmanager350 and CB servers380nhave failed. As explained above, this can be accomplished using any failure detection technique known in the art. In case CB server380ndetects a failure of anApp server390nbefore meetingmanager350, CB server380nnotifiesprocess manager311 through thezone manager313 communication path. In some embodiments, the zone managers communicate with each other and the designated ZM/GK314 using the well-known TCP/IP protocol and simple messages whose content and format are readily apparent to those of ordinary skill in the inter-process communication arts. In other embodiments, the WebEx Transport Layer protocol is used.
• The WebEx Transport Layer protocol (TP) is responsible for providing point-to-point connectivity between a WebEx client and the WebEx server. The TP layer will attempt to create direct TCP connections and use TCP to communicate between the client and server. For clients that sit behind firewalls, particularly for those that are unable to create direct TCP connections, the WebEx TP layer will automatically create virtual sockets based upon HTTP. This enables the client to communicate with the server through most firewalls.[0080]
• Since the HTTP protocol functions on a Request/Response basis, it is always the client that issues the Request command. Hence, in order to provide a bi-directional communication channel, the client actively polls the server in order to fetch the data that may be sent from the server to the client. The details of this implementation are available in the copending and commonly-assigned Application for a U.S. patent Ser. No. 09/449,011, filed on Nov. 24, 1999, “Emulating a Persistent Connection Using HTTP,” cited and incorporated above.[0081]
• If a failure of[0082]App server390nis detected,operation900 proceeds to stage920. Otherwise stage910 is repeated. Instage920,meeting manager350 places any CB servers380nconnected to failedApp server390nin a suspend state and receives a request for anew App server390nfrom CB server380ninstage930.Meeting manager350 then requests thatprocess manager311 launch anew App server390n(stage940).Process manager311 launches thenew App server390nand notifies meeting manager350 (stage950).
• Once meeting[0083]manager350 has received notification that thenew App server390nhas been launched,meeting manager350 resumes (i.e., removes from the suspend state) CB server380nand connects it to thenew App server390n. (App server state is restored from a backup meeting manager, through any of a number of standard and common means well-known in the art.) Meeting manager continues to monitor the status ofApp server390n(stage910). Note that all logical server-to-logical server and logical server-to-PM communications employZMs313 and314.
• FIG. 10 illustrates the license/log manager failure detection and[0084]recovery operation1000. First,meeting manager350 checks whetherlicense manager360 orlog server370 have failed, using similar techniques to the ones described above in reference to FIGS. 8 and 9. If a failure is detected,operation1000 proceeds tostage1020. Otherwise,stage1010 is repeated until a failure is detected.Meeting manager350, in turn, sends a request to processmanager311 to launch anew license manager360 or log server370 (stage1020), as required.Stage1030 then determines whether thenew license manager360 orlog server370 has successfully come on-line, in whichcase meeting manager350 continues to monitor the status oflicense manager360 and log server370 (stage1010). Otherwise, stages1030 and1040 are repeated until anew license manager360 orlog server370 has been successfully started.
• Note that the reliable TP layer keeps all data and resends/reloads it into the replacement license and/or log server as needed.[0085]
• FIGS.[0086]8-10 thus show howmeeting manager350 monitors the status of other components in its meeting zone310n. However, to provide even more effective fault tolerance, the status ofmeeting manager350 must also be monitored to prevent a single point of failure in the system. This is accomplished by providing both a primary and one or morestandby meeting managers350 in each meeting zone310n. In addition,process manager311 is responsible for detecting failure of theprimary meeting manager350 and transferring control to one of thebackup meeting managers350. Operability of the process manager, in turn, is guaranteed by a hardware time-out restart process.
• FIG. 11 illustrates meeting manager failure detection and[0087]recovery operation1100. In each meeting zone310n(referring to FIG. 3), there is instantiated oneprimary meeting manager350 and one or more secondary meeting managers (not shown).Process manager311 continually checks whetherprimary meeting manager350 has failed (stage1110), again using standard failure detection techniques. If a failure ofprimary meeting manager350 is in fact detected,operation1110 proceeds tostage1120. Otherwise,stage1110 is repeated.
• In[0088]stage1120,process manager311 launches a new standby meeting manager. The pre-existing standby meeting managers, advised of the failure of primary meeting manager byprocess manager311, elect (through any of several well-known server election or promotion mechanisms) one of their own (step1140) to take over as primary and broadcast an election message (stage1140). One of the standby meeting managers is thus selected as the new primary meeting manager350 (stage1150). In the event only one standby MM is presently configured, the election message ofstage1140 is simply construed as a command to become the primary MM.
• The standby meeting manager(s)[0089]350, CB servers380n,App server390n,ping servers355,license manager360, andlog server370 in the same meeting zone310nas newprimary meeting manager350 connect to new primary meeting manager350 (stage1160) and register with it (stage1170) so that the new primary meeting manager can continue to monitor the status of these servers. Newprimary meeting manager350 recovers its server state (stage1180) and receives reports from CB servers380non the status of any active conferences handled by CB servers380n(stage1190). Finally, newprimary meeting manager350 recovers meeting information for all meetings handled in the meeting zone310n(stage1190).Process manager311 monitors the status of new primary meeting manager350 (stage1110).
• CB server[0090]380ninterfaces withclient browser320 through application protocol entities (APEs) joined to agent sessions. FIG. 12 is a block diagram illustrating the software components ofclient computers210nandserver computers220n(FIGS. 2A and 2B) involved in the communications between CB server380nandclient browser320. In particular, communications channels are established between transaction processing (TP)server1250 and Application Resource Manager (ARM)server1240 onserver computer220nandTP client1230 andARM client1220 onclient computer210n. Thus,conference manager1260 andApp server390n(both logically part of CB server380n) communicate withclient computer210nvia the communication channels maintained byARM server1240 andTP server1250.
• FIGS.[0091]13A-13C are flow diagrams illustrating theoperation1300 of CB server380nandApp server390nto setup communications with client browser320 (FIG. 3). First, CB server380ncreates an agent session (stage1305). The agent session controls communications fromclient computer210nto CB server380nand can launch new, additional data sessions if required. To communicate with CB server380n,client computer210n, in turn, creates an APE (stage1310) and joins the APE to the agent session (stage1315). In stage1316, CB server380nsends a list of all existing session to theclient computer210n; instage1317, the client must chose whether to join all or only some sessions. Ifclient computer210njoins all sessions, control passes to stage1320, shown in FIG. 13B. If not, stage1318, the client joins only selected sessions before control passes tostage1320.
• Stage[0092]1320 (FIG. 13B) determines whether the user ofclient computer210nhas elected to create a new session (e.g., to share an application), in whichcase operation1300 proceeds tostage1325. Otherwise,operation1300 proceeds tostage1360.Client computer210nAPE then sends a message to the agent session APE of CB server380nrequesting a new session (stage1325). CB server380n, in turn, requests a new session fromApp server390n(stage1330) andApp server390ncreates the new session for the conference (stage1335).App server390nalso creates a new APE and joins the new session to the new APE (stage1340). CB server380n, in turn, sends the new session's ID toclient computer210n(stage1345).Client computer210nlaunches an application (stage1350), creates a new APE for the application and joins the new APE to the new session (stage1355, referring to FIG. 13C).
• [0093]Stage1360 determines if anew client computer210nwants to join an existing session, in whichcase operation1300 proceeds tostage1370. Otherwise,operation1300 terminates.Client computer210nrequests joining the session (stage1370), concludingoperation1300.
• FIG. 14 is a block diagram illustrating the communication channels established between[0094]client computers210A and210B during an on-line conference, in accordance with an embodiment of the invention.Client computer210A connects toCB server380B in meetingzone310A viaARM server1240 andTP server1250. In addition,CB server380B established a high-speed real-time messaging link1420 withCB server380C in meetingzone310B using a real-time messaging service (RTMS)1410. In one embodiment of the present invention,RTMS1410 is implemented using the well-known TCP/IP communications protocol. In some alternate embodiments, the WebEx Transport Protocol, discussed above, is used.
• [0095]CB server380C, in turn, connects toclient computer210B via itsown ARM server1240 and TP server1250 (not shown).
• FIG. 15 is a flow diagram of[0096]operation1500 for transmitting data fromclient computer210A toclient computer210B using distributed collaborative computer system300 (FIG. 3). First,CB server380B establishes a link toCB server380C using real-time messaging service1410 (stage1510, as illustrated in FIG. 14). The session information is then replicated fromCB server380B toCB server380C (stage1520). Data routed fromclient computer210A is then transmitted fromCB server380B toCB server380C over real-time messaging service1410 (stage1530). The data received byCB server380C is then routed toclient computer210B using TP server1250 (stage1540).Stage1550 then determines if additional data needs to be transmitted fromclient computers210A and210B, in which case stages1530-1550 are repeated. Otherwise,operation1500 terminates.
• Distributed[0097]collaborative computer system300 allows users ofclient computers210nto participate in on-line conferences by sharing both audio and video signals. In particular, distributedcollaborative computer system300 allows users to share images of a document that can be marked-up by conference participants (document viewing). Document viewing is described in further detail in U.S. Pat. No. 5,577,188 “Method to Provide for Virtual Screen Overlay” and co-pending and commonly-assigned U.S. patent application Ser. Nos. 09/471,938 and 09/591,377 (filed on Dec. 23, 1999 and Jun. 9, 2000, respectively), cited and incorporated above. In addition, users may share control of an application program executed on any of theclient computers210nparticipating in the on-line conference (a process known as application sharing). Application sharing is described in further detail in co-pending and commonly-assigned U.S. patent application Ser. No. 09/442,424 (filed Nov. 17, 1999), cited and incorporated above.
• During document viewing, the presenter may choose to skip one or more pages in the document being viewed. FIGS. 16A and 16B are flow diagram illustrating the[0098]skip page operation1600 used to control transmission of pages between the presenter'sclient computer210nand other participants'client computers210n.
• First, an[0099]App server390nproviding the document viewing application (also referred to as the docview server) assigns unique IDs to each page in the document being viewed (stage1605, FIG. 16A). The page IDs and page content data are then passed toARM client1220 and fromARM client1220 to ARM server1240 (stage1610).ARM server1240, in turn, begins transmitting the document page IDs and data over a shared data queue on high-speed real-time messaging link1420 (stage1615). The first page ID is then sent to allclient computers210nconnected to the conference (stage1620).Client computers210n, in turn, request the first page data from the shared data queue (stage1625) and CB server380nsends the first page data toclient computers210n(stage1630).Stage1635 then determines whether the presenter has elected to jump to a new page in the shared document, in whichcase operation1600 proceeds tostage1640. Otherwise,operation1600 proceeds tostage1655. In stage1640 (FIG. 16B), the presenter'sclient computer210nbroadcasts the new page ID to allclient computers210nparticipating in the conference. The new page data is then transmitted over the shared data queue (stage1645) andclient computers210nrequest the new page from the shared media queue (stage1650).
• Alternatively,[0100]stage1655 determines if all data transmitted on the shared data queue has been received, in which case the docview server is notified (stage1660. Otherwise,operation1600 proceeds tostage1635.
• [0101]Stage1665, in turn, determines whether the shared data queue is no longer needed, in which case the shared data queue is emptied (stage1670) andoperation1600 terminates. Otherwise,operation1600 proceeds tostage1635.
• FIG. 17 is a flow diagram of a[0102]client browser operation1700, in accordance with some embodiments of the invention. First,client browser320 receives conference parameters from CB server380n(stage1710).Client browser320 then connects to CB server380n(stage1720) to participate in the conference.Stage1730 checks the status of CB server380n. If a failure of CB server380nis detected,client browser320 attempts to reconnect to a new CB server380n(stage1740) and stages1710-1730 are repeated. Otherwise,client browser320 continues to monitor the status of CB server380n.
• FIGS. 18A, 18B,[0103]18C1-3,19A,19B,20A,20B and20C are views of web pages displayed by client browser320 (FIG. 3) during operation of distributedcollaborative computer system300.
• Meeting center web page[0104]1800 (FIGS. 18A, 18B and18C1-3) is displayed when a user first accesses web server335 (FIG. 3) throughclient browser320. Meetingcenter web page1800 contains a list of current and scheduled meetings the user may want to join. In addition, the user may create a new meeting by selecting createmeeting button1810, causing a sign in prompt to be displayed in meeting center web page1800 (FIG. 18B). If the user is not already registered with the service, the user can register by selectingnew user link1820. Otherwise, the user can enter ID and password information inlogin prompt1830. If the user's data is successfully authenticated with the information stored inweb database337 and/or central operation database340 (FIG. 3), a create new meeting prompt1840 is displayed in meeting center web page1800 (FIGS.18C1-3). The user can then enter meeting parameters such as date, time, and attendee list by filling innew meeting prompt1840. The user can also edit meeting options by selectingedit options button1850, thereby causing meeting options web page1900 (FIGS.19A-19B) to be displayed. Once the user has entered the desired meeting information on meetingcenter web page1800, the user can either schedule the meeting by pressingschedule button1860 or start the meeting by pressing start nowbutton1870.
• Meeting[0105]options web page1900 allows the user to set specific meeting options such as features, client type, frequency and reminders. Once the user is satisfied with the selected options, the user can return to meetingcenter web page1800 by pressing submitbutton1910.
• Meeting web page[0106]2000 (FIGS.20A-20C) is displayed to the user during a meeting.Meeting web page2000 includes a sharedpane2010, anattendee pane2020 and amessage pane2030. Information shared among meeting participants are displayed in sharedpane2010. The user can share images, documents, applications, web pages, desktops and whiteboards by selecting an appropriate entry from tools menu2040 (FIG. 20B). For example, if the user selects to share an image to be marked up by the meeting participants, the image is displayed in shared pane2010 (FIG. 20C). One or more users can then mark up the image by selecting a drawing tool from drawingmenu2050 and drawing over the image.Attendee pane2020 contains a list of meeting attendees. Alternatively,attendee pane2020 can used to display polls taken among the meeting attendees or a video conferencing images. Finally,message pane2030 can used to compose, send and receive messages among two or more meeting attendees.
• Since conference information is replicated across all CB servers[0107]380nhandling the conference and can be reconstructed by meetingmanager350, failure of one or more CB servers380ndoes not disrupt the conference and can be gracefully recovered. As a result, the distributed collaborative computing system of the present invention eliminates the single point of failure limitation of prior art collaborative computing systems. In addition, sincemultiple server computers220nare used to handle an on-line conference, the distributed collaborative computing system of the present invention may handle conferences with an arbitrary number of participants, without any limitations imposed by the processing capacity of any single server computer. By contrast, prior art systems were limited to conferences whose participants could all be handled by a single server computer.
• Alternate Embodiments[0108]
• The order in which the steps of the present method are performed is purely illustrative in nature. In fact, the steps can be performed in any order or in parallel, unless otherwise indicated by the present disclosure.[0109]
• The method of the present invention may be performed in either hardware, software, or any combination thereof, as those terms are currently known in the art. In particular, the present method may be carried out by software, firmware, or microcode operating on a computer or computers of any type. Additionally, software embodying the present invention may comprise computer instructions in any form (e.g., source code, object code, interpreted code, etc.) stored in any computer-readable medium (e.g., ROM, RAM, magnetic media, punched tape or card, compact disc (CD) in any form, DVD, etc.). Furthermore, such software may also be in the form of a computer data signal embodied in a carrier wave, such as that found within the well-known Web pages transferred among computers connected to the Internet. Accordingly, the present invention is not limited to any particular platform, unless specifically stated otherwise in the present disclosure.[0110]
• While particular embodiments of the present invention have been shown and described, it will be apparent to those skilled in the art that changes and modifications may be made without departing from this invention in its broader aspect and, therefore, the appended claims are to encompass within their scope all such changes and modifications as fall within the true spirit of this invention.[0111]

  

  

Claims (16)

We claim:

1. A method of distributed collaborative computing comprising:

partitioning a collaboration function into sub-functions;

assigning at least one said sub-function to each of a plurality of logical processes;

associating a respective management process with each of said plurality of logical processes, said logical processes configured so that each said logical process is capable of communicating with every other said logical process thru said respective management process;

communicating between said logical processes using said respective management processes; and

monitoring said respective management processes with a single supervisor process;

wherein said monitoring further comprises re-creating one or more said logical processes in response to detecting a failure of one or more said logical processes.

2. The method ofclaim 1, said detecting further comprising:

monitoring a message stream to determine a responsiveness of said logical process; and

if said responsiveness ceases, signaling said failure to said single supervisor process.

3. The method ofclaim 1, said re-creating further comprising:

spawning a new logical process;

assigning to said new logical process said at least one sub-function corresponding to said failed logical process;

recovering the state of said failed logical process into said new logical process; and

associating a new management process with said new logical process.

4. The method ofclaim 1, said re-creating further comprising:

activating a stand-by logical process, said stand-by process having been instantiated prior to said detecting;

assigning to said stand-by logical process said at least one sub-function corresponding to said failed logical process;

recovering the state of said failed logical process into said new logical process;

associating a new management process with said stand-by logical process; and

spawning a new stand-by logical process.

5. A computer program for use in distributed collaborative computing, comprising computer instructions for:

partitioning a collaboration function into sub-functions;

assigning at least one said sub-function to each of a plurality of logical processes;

associating a respective management process with each of said plurality of logical processes, said logical processes configured so that each said logical process is capable of communicating with every other said logical process thru said respective management process;

communicating between said logical processes using said respective management processes; and

monitoring said respective management processes with a single supervisor process;

wherein said monitoring further comprises re-creating one or more said logical processes in response to detecting a failure of one or more said logical processes.

6. The computer program ofclaim 5, said detecting further comprising:

monitoring a message stream to determine a responsiveness of said logical process; and

if said responsiveness ceases, signaling said failure to said single supervisor process.

7. The computer program ofclaim 5, said re-creating further comprising:

spawning a new logical process;

assigning to said new logical process said at least one sub-function corresponding to said failed logical process;

recovering the state of said failed logical process into said new logical process; and

associating a new management process with said new logical process.

8. The computer program ofclaim 5, said re-creating further comprising:

activating a stand-by logical process, said stand-by process having been instantiated prior to said detecting;

assigning to said stand-by logical process said at least one sub-function corresponding to said failed logical process;

recovering the state of said failed logical process into said new logical process;

associating a new management process with said stand-by logical process; and

spawning a new stand-by logical process.

9. A computer-readable medium storing a computer program executable by a plurality of server computers, the computer program comprising computer instructions for:

partitioning a collaboration function into sub-functions;

assigning at least one said sub-function to each of a plurality of logical processes;

associating a respective management process with each of said plurality of logical processes, said logical processes configured so that each said logical process is capable of communicating with every other said logical process thru said respective management process;

communicating between said logical processes using said respective management processes; and

monitoring said respective management processes with a single supervisor process;

wherein said monitoring further comprises re-creating one or more said logical processes in response to detecting a failure of one or more said logical processes.

10. The computer-readable medium ofclaim 9, said detecting further comprising:

monitoring a message stream to determine a responsiveness of said logical process; and

if said responsiveness ceases, signaling said failure to said single supervisor process.

11. The computer-readable medium ofclaim 9, said re-creating further comprising:

spawning a new logical process;

assigning to said new logical process said at least one sub-function corresponding to said failed logical process;

recovering the state of said failed logical process into said new logical process; and

associating a new management process with said new logical process.

12. The computer-readable medium ofclaim 9, said re-creating further comprising:

activating a stand-by logical process, said stand-by process having been instantiated prior to said detecting;

assigning to said stand-by logical process said at least one sub-function corresponding to said failed logical process;

recovering the state of said failed logical process into said new logical process;

associating a new management process with said stand-by logical process; and

spawning a new stand-by logical process.

13. A computer data signal embodied in a carrier wave, comprising computer instructions for:

partitioning a collaboration function into sub-functions;

assigning at least one said sub-function to each of a plurality of logical processes;

associating a respective management process with each of said plurality of logical processes, said logical processes configured so that each said logical process is capable of communicating with every other said logical process thru said respective management process;

communicating between said logical processes using said respective management processes; and

monitoring said respective management processes with a single supervisor process;

wherein said monitoring further comprises re-creating one or more said logical processes in response to detecting a failure of one or more said logical processes.

14. The computer data signal ofclaim 13, said detecting further comprising:

monitoring a message stream to determine a responsiveness of said logical process; and

if said responsiveness ceases, signaling said failure to said single supervisor process.

15. The computer data signal ofclaim 13, said re-creating further comprising:

spawning a new logical process;

assigning to said new logical process said at least one sub-function corresponding to said failed logical process;

recovering the state of said failed logical process into said new logical process; and

associating a new management process with said new logical process.

16. The computer data signal ofclaim 13, said re-creating further comprising:

activating a stand-by logical process, said stand-by process having been instantiated prior to said detecting;

assigning to said stand-by logical process said at least one sub-function corresponding to said failed logical process;

recovering the state of said failed logical process into said new logical process;

associating a new management process with said stand-by logical process; and

spawning a new stand-by logical process.

Images