Electrical computers and digital processing systems: multicomput – Distributed data processing – Client/server
Reexamination Certificate
2000-05-30
2004-02-03
Maung, Zarni (Department: 2154)
Electrical computers and digital processing systems: multicomput
Distributed data processing
Client/server
C709S217000, C709S219000, C709S220000, C709S241000, C709S241000, C709S241000, C370S486000, C370S487000
Reexamination Certificate
active
06687735
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates generally to the field of software applications that are distributed over an information network, and specifically to the dynamic distribution and division of software applications and other associated components between the client device(s) and server(s) on the network.
2. Description of Related Technology
Client-server network architectures are well known in the data networking arts. Often, there is a need to divide the software associated with such client-server systems into multiple components which are distributed across one or more devices of the network. A Distributed Application (DA) is a computer program that is broken into such multiple components. These components are then distributed as client and server portions of the distributed application; hereinafter known as the Distributed Application Client Portion (DACP) and the Distributed Application Server Portion (DASP).
Typically, client devices, hereinafter known as “the client”, will have no where near the processing power of a server machine (“the server”). Examples of client devices include set-top boxes and networked personal computers (PCs). The DACP will be considered “thin” meaning that it will generally have minimal functionality so as to minimize the resource burden placed on the client device. Therefore, most of the distributed application will run on the server, the configuration of which is selected so as to be capable of handling the needs of multiple DACPs simultaneously. Hence, a single DASP can handle processing for multiple DACPs for the same application. In addition, a single server can run as many DASPs as its hardware and operating system constraints will allow. A server may run multiple DASPs for the same or different applications simultaneously.
In general, the server downloads the DACP to the client device(s) upon the occurrence of a specified event, such as the user selecting a function on their set-top box. The author of the distributed application software determines how processing is divided between the DASP and the DACP. The DACP is, for example, often limited to handling only input/output (I/O) and graphics functionality on the client device. The DACP places consumer input data into messages and sends them to the distributed application server for processing.
FIG. 1
demonstrates the foregoing distributed application model for a single client to server relationship.
With the advent of digital cable systems that provide downloadable software applications and reverse communications with entities such as multi-system operator (MSO) head-ends and the Internet, set-top box and client device resource requirements have risen dramatically. Downloadable software applications taking full advantage of resources defined in standards such as the OpenCable Standard, ATSC (Advanced Television Systems Committee), DASE (Digital television Applications Software Environment) and DVB (Digital Video Broadcasting) can demand considerable CPU, memory, and storage resources, which leads to greater set-top box complexity and cost.
In order to allow for the operation of such resource-intensive software applications while using only the minimum profile set-top equipment, a technique is needed to dynamically off-load portions of the application execution to server processes not residing in the client device. Existing prior art distributed applications do not permit such dynamic allocation or “scaling” between the server and client portions, especially based on upon the client device configuration, thereby requiring careful consideration of the client device configuration during development of the distributed application. For example, the well-known prior art CORBA (Common Object Request Broker Architecture) environment is a non-scaling three tiered system. The three tiers of the CORBA system consist of a user interface (UI) tier, computational processing tier, and database tier. The client device employs a graphical user interface (GUI) as the UI tier. One or more servers containing the processing tier and the database tier are also employed in this system. This three tier approach does not, however, provide any facility for scaling of the distributed application between the server(s) and client device(s), whether based on client device configuration or otherwise. Specifically, such three tier systems do not define moving objects around dynamically at runtime, in order to make a client “thinner” or “fatter” based on the resource capability of the client.
Similarly, the well known COM
+
and DCOM systems produced by Microsoft Corporation provide templates or frameworks for the server and client pieces, yet these frameworks are not automatically self-distributing and do not provide for distributed application scaling or dynamic movement of objects based on resource capability at runtime.
In addition to distributed application scaling and distribution described above, a technique is needed whereby server portions of such applications can be dynamically distributed among all of the available servers at the head-end, thereby obviating constant manual configuration by the MSO, and permitting the head-end application management to be self-governing. Prior art distributed application systems typically accomplish server-end load balancing by measuring communications bandwidth or similar parameters representative of the loading on a given server, and re-allocating the load accordingly. Hence, these prior art systems conduct load balancing at the lower layers (e.g., physical or transport layers of the OSI model) of the protocol stack. This generic network approach works well for PC network topologies, yet is not well suited to a cable (e.g., HFC) network which requires a finer grained approach.
SUMMARY OF THE INVENTION
The present invention satisfies the aforementioned needs by providing an improved method and apparatus for balancing and dynamically distributing the various components of a distributed application within a client-server environment.
In a first aspect of the invention, an improved method of starting and downloading the client portion of a distributed application on a client-server network is disclosed. The method generally comprises the steps of providing a distributed application having a plurality of separable modules; disposing the distributed application on a server within the network; transferring a first portion of the plurality of modules to a client device via a first communication channel; attempting to establish a second “reverse” communication channel between the client device and the server; detecting when the reverse channel is not established, and subsequently transferring additional modules of the distributed application to the client device from said server in a predetermined increment until the second communication channel can be successfully established. In one exemplary embodiment, a number of different secondary communication channel options including a cable modem (DOCSIS), out-of-band (OOB), and telephone company (telco) modem, are used hierarchically such that if the first selected channel in the hierarchy can not be established, the second is next tried, and so forth. Downloading of the communications modules associated with these channels is also performed hierarchically so that only the minimum portion of the distributed application necessary to facilitate communication via the reverse channel is downloaded to the client device.
In a second aspect of the invention, an improved server/head-end MSO architecture adapted for distributed application load balancing is disclosed. The improved head-end architecture generally comprises a plurality of servers arranged in a server farm; a distributed application comprising at least one server portion disposed on at least one of the servers in the farm; a runtime entity comprising a plurality of software processes which facilitate the allocation of client device load between the various servers in the farm; and a message protocol adapted for communication be
Ladd Patrick
Logston Gary
Gazdzinski & Associates
Hu Jinsong
Maung Zarni
Tranceive Technologies, Inc.
LandOfFree
Method and apparatus for balancing distributed applications 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 balancing distributed applications, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and apparatus for balancing distributed applications will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3342840