Electrical computers and digital processing systems: multicomput – Computer-to-computer session/connection establishing – Network resources access controlling
Reexamination Certificate
1997-12-08
2003-05-20
Geckil, Mehmet B. (Department: 2142)
Electrical computers and digital processing systems: multicomput
Computer-to-computer session/connection establishing
Network resources access controlling
C709S219000, C709S225000, C709S227000, C707S793000, C713S152000
Reexamination Certificate
active
06567853
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates generally to data transfer in an input/output (I/O) system, and more particularly to data transfers that utilize non-server reconnections from storage devices.
BACKGROUND OF THE INVENTION
Current computing environments typically comprise computer networks. Whether locally connected, or connected via a remote link, such as through a dial-in modem link, computer systems normally communicate via a server device. These computer systems, i.e., clients, require performance of various services, while the server device, i.e., servers, are the hardware/software network components that perform these services. Included among these services are electronic mail, file transfers, and remote database access applications. Moving data between computers and between processes can result in a large amount of computing overhead for servers, especially when data is moved to different locations in a server's local storage, such as onto a storage device.
Typically, a server masks the appearances of storage devices from the client. Thus, a client must make a data request of a server in file name or other terms with the server mapping the request to one or more of its attached storage devices. Storage device interface protocols, such as device address, tracks, and sectors, are therefore not usually used between the client and server. In order to alleviate some of the overhead in the server, including reducing the cost in terms of the server's resources of memory, data paths, and transfer bandwidth, storage server systems seek a design in which storage data may be directly transferred between clients acting as requesting systems and the storage devices, rather than being transferred through a server system. While alleviating some of the overhead, a further benefit of the design is that storage capacity may be added without requiring an increase in the size of the server, thus providing greater storage scalability without a concomitant scaling of the server's resources.
While direct client system-storage device transfer may avoid scaling up of the transfer resources in servers, unfortunately, increased I/O communication overhead results. A client system must both communicate with a server and storage devices, and the server must have additional communications with clients and storage for each request to manage and protect its device and data resources. Scalability, therefore, is advantageous primarily where the amount of data transferred per request is large, such as in file transfer. Further, a design for scalability should allow for future direct network attachment of storage devices. Also, the “open” nature of the desirable client access requires that servers be able to manage and restrict access to storage devices by client systems, permitting only that access needed for each request. In addition, if transfer is to or from more than one storage device, the client must deal with data in parts in handling data transfer to or from the several storage devices for a single server request.
Lawrence Livermore National Labs (LLNL) provides an example of an attempt to achieve a scalable I/O system, i.e., to be able to have large amounts of storage/peripherals, DASDs (direct access storage devices) in particular, without requiring that servers have the processing, memory buffer, and data transfer rate capacity to pass all client-requested data through the server. For LLNL, a read-write with ticket (RWT) approach provides a general method for prevalidating requests from client systems to DASD and using digital signatures. Unfortunately, using digital signatures results in potential synonyms and increases complexity to DASDs by requiring validation of the signature. In general, robust digital signatures are long, thus requiring more device storage for validated pending requests. Further, RWT requires that data extent address information be returned to the client system, thus potentially allowing a successfully forged signature to be created and sent with a DASD command to a DASD device. LLNL RWT also requires explicit post-data transfer server communication to cancel the ticket in the DASD. In addition, RWT requires that the DASD a priori know the network address of the client system.
A need exists for a method and system for achieving a scalable input/output system that provides a “trusted” server to device control connection and protocol for the server to set up limited access transfer parameters for clients.
SUMMARY OF THE INVENTION
The present invention meets these needs provides a method and system for a scalable I/O system. The scalable I/O system includes a server, at least one client, and at least one storage device. The server interfaces with the at least one client and at least one storage device. The at least one storage device and at least one client also interface. The server initiates data transfer from the storage device on behalf of an open client (i.e., a client not closed within a fixed system or set of systems). The server further sets up a disconnect state in the at least one storage device to be reconnected for transfer to a non-server interface. The server further passes information to the open client that is requesting data transfer, which allows the open client to determine dimensions of data transfer, number of storage devices that require accessing for the data transfer, and the relationship of the data transfer of each storage device to the original request sent to the server.
Through the present invention, scalable growth of storage on a server or servers results without requiring comparable growth in server resources, e.g., memory for data buffers, data transfer bus bandwidth, etc. Further, access to the storages directly from clients via networks or conventional storage interfaces is achieved without requiring clients to a priori understand storage data locations or storage data address parameters. Additionally, the present invention provides security cost and performance effectiveness for storage devices and storage systems. Neither encryption nor Kerberos authentication is required, nor does it require that the storages act in “channel mode” as I/O or network communication initiators. Transfers are able to be accomplished with a minimum of inter-unit communication overhead, and storage device operations are able to begin earlier in a sequence, with access operations overlapped with some server-client communication. These and other advantages of the aspects of the present invention will be more fully understood in conjunction with the following detailed description and accompanying drawings.
REFERENCES:
patent: 4649473 (1987-03-01), Hammer et al.
patent: 5204954 (1993-04-01), Hammer et al.
patent: 5212784 (1993-05-01), Sparks
patent: 5404527 (1995-04-01), Irwin et al.
patent: 5566331 (1996-10-01), Irwin, Jr. et al.
patent: 5584042 (1996-12-01), Cormier et al.
patent: 5590334 (1996-12-01), Saulpaugh et al.
patent: 5613124 (1997-03-01), Atkinson et al.
patent: 5613155 (1997-03-01), Baldiga et al.
patent: 5694615 (1997-12-01), Thapar et al.
patent: 5793965 (1998-08-01), Vanderbilt et al.
patent: 5835724 (1998-11-01), Smith
patent: 5838916 (1998-11-01), Domenikos et al.
patent: 5857188 (1999-01-01), Douglas
patent: 5862339 (1999-01-01), Bonnaure et al.
patent: 5867661 (1999-02-01), Bittinger et al.
patent: 5893116 (1999-04-01), Simmonds et al.
patent: 5893918 (1999-04-01), Beardsley et al.
patent: 5903733 (1999-05-01), Hong
patent: 5905872 (1999-05-01), DeSimone et al.
patent: 5915093 (1999-06-01), Berlin et al.
patent: 5940841 (1999-08-01), Schmuck et al.
patent: 5996014 (1999-11-01), Uchihori et al.
patent: 6003084 (1999-12-01), Green et al.
patent: 6006268 (1999-12-01), Coile et al.
patent: 6014702 (2000-01-01), King et al.
patent: 6023698 (2000-02-01), Lavey, Jr. et al.
patent: 6029175 (2000-02-01), Chow et al.
patent: 6032224 (2000-02-01), Blumenau
patent: 6088451 (2000-07-01), He et al.
patent: 6092196 (2000-07-01), Reiche
patent: 6108703 (2000-08-01), Leighton et al.
patent: 6192408 (2001-02-01), Vahalia et al
Geckil Mehmet B.
Sawyer Law Group LLP
Vaughn, Jr. William C.
LandOfFree
Scalable I/O system for the efficient transfer of storage... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Scalable I/O system for the efficient transfer of storage..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Scalable I/O system for the efficient transfer of storage... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3002798