AU2007200468A1 - High speed information processing and mass storage system and method, particularly for information and application servers - Google Patents

Description

P/00/011 Regulation 3.2

AUSTRALIA

Patents Act 1990 COMPLETE SPECIFICATION STANDARD PATENT Invention Title: High speed information processing and mass storage system and method, particularly for information and application servers The following statement is a full description of this invention, including the best method of performing it known to us: HIGH SPEED INFORMATION PROCESSING AND MASS STORAGE SYSTEM AND METHOD, PARTICULARLY FOR C~1 INFORMATION AND APPLICATION SERVERS 00 Related Applications: This application is related to Patent Application No. PCT1US99/05231I of Richard Dellacona filed March 10, 1999, which is based upon Provisional Patent Application Serial No. 60/077,643, filed March 10, 1998, and to U.S. Patent Application Serial No. 09/071282 of Richard Dellacona filed May 1, 1998, all of which are hereby incorporated herein by reference.

BACKGROUND) OF THE INVENTION Field of the Invention This invention relates to a high speed, microcomputer based, Fibre Channel compatible and fault tolerant information processing and mass storage system especially suited for information servers and application servers. In particular, the present invention relates-to a method and apparatus for information processing and storage involving a unique and extremely versatile system architecture, including a dual loop arbitrated, Fibre Channel capable, multiple-fault tolerant, hot-swappable .mass storage disk array and including a method and apparatus for providing an enterprise-wide information or application server system using such disk array.

State of the Art Efforts have been made in the past to provide a mass storage file server capable of delivering information throughout an enterprise with high speed data throughput, scalable data storage capability in a convenient, easily configurable enclosure using well known, industry standard operating software and hardware.

However, such systems have typically experienced many shortcomings and problems associated with equipment incompatibility as well as with the inability of presently available computer and communications hardware to sustain performance and service failure of component devices. Such shortcomings have included the lack of capability to add storage devices to accommodate increased storage requirements or L to replace failed storage devices without the need to completely power down the Cl i.:3rmat-on;server. Some of the compatibility problems have involved, for cxample, bottlenecks in sharing information among the equipment components of various vendors. The above-referenced Dellacona patent applications address some of these S 5 problems and others" and provide unique solutions which are described and claimed therein.

00 IND The present invention further addresses some of the problems discussed in the C) referenced Dellacona applications, as well as others. For example, the present invention further addresses the problem of scalability and customization. in S 10 information processing and storage systems for different applications. In some applications, there may be a greater need for processing capability rather than storage capacity, while other applications may require just the opposite. Yet other applications may require storage expansion for existing information processing systems. This invention provides an architecture which will readily accommodate such needs.

In addition, mnass storage systems can create considerable heat, particularly where they are disk drive based. If the heat is not effectively removed, it can affect the reliability and life of the system. Often, it is difficult to remove the heat because of obstructions caused by the physical configuration of back planes and mid planes which act as barriers to air flow. Typically, for example, all of the disk drives of a mass storage module or array are typically plugged into connectors on the face of a backplane or mid plane that extends across the entire module. Whether enclosed in a cabinet or other enclosure or rack mounted without an enclosure, air flow through the module is inhibited by this type of structural arrangement, and excessive heating can occur, particularly in the vicinity of the disk drives.

Also, it is desirable to be able to hot swap individual disk drives of a mass storage module to accommodate the need for more storage capacity, but the system storage requirements may outgrow the capacity of the module and it may also be desirable to have the capability of adding modules without powering down the system. Of course, this capability must be provided without disturbing the operation of the existing module and with a minimum of signal degradation as modules are added.

004880335 3 It will be understood that any discussion of prior art herein does not constitute an admission as to the common general knowledge of a person skilled in the art.

SSUMMARY OF THE INVENTION Advantageously, embodiments of the present invention will address one or more of the O0 5 foregoing problems and/or shortcomings of the prior art through the provision of an information processing and mass storage method and system, including a unique mass storage array, particularly suited for information servers or application servers, with a novel system architecture which permits the addition or replacement of storage devices without interrupting or seriously degrading the operation of the system and which is highly fault-tolerant and reliable.

Embodiments of the invention may also address one or more of the foregoing problems by providing a novel physical layout that permits the effective removal of heat from a system module containing heat creating components.

In accordance with one embodiment of the invention, there is provided a high speed mass storage system which is readily expandable to increase its storage capacity while the system is in operation comprising: a first server including a first controller and at least one CPU; a second server including a second controller and at least one CPU; first and second mass storage modules each including: a plurality of plug-in storage devices for storing information; a storage device bypass circuit board associated with each storage device, each storage device being plugged into a connector on the storage device bypass circuit board; a module bypass circuit board including an optical input/output connector for outputting electrical signals from the module as light signals and for inputting light signals into the module as electrical signals, and wherein the optical input/output connectors of the module bypass circuit boards of the first and second mass storage modules are connected by a fiber optic transmission medium such that signals are communicated between the first and second mass storage modules in the form of light; 004880335 3a said first controller providing a communication path between the first server and each ,I said storage device through its associated storage device bypass circuit board and through the Smodule bypass circuit board, said second controller providing a communication path between the Ssecond server and each said storage device through its associated storage device bypass circuit 0 5 board and the module bypass circuit board; and 00 at least one of said servers being operative to establish direct communication between the first and second controllers, and said first and second controllers being operative to maintain direct communication between the first and second controllers independent of said at least one SCPU of said first server and said at least one CPU of said second server.

In accordance with another embodiment of the invention there is provided a high speed mass storage system adapted to be readily expandable to increase its capacity while the system is in operation comprising: a first server including a first controller and at least one CPU; a second server including a second controller and at least one CPU; first and second mass storage modules each including: a plurality of plug-in disk drives for storing information; a disk drive bypass circuit board associated with each disk drive and including a disk drive connector at one edge thereof and a bypass board connector at another edge thereof, each disk drive being plugged into said disk drive connector on the disk drive bypass circuit board; a module bypass circuit board including an optical input/output connector for outputting electrical signals from the module as light signals and for inputting light signals into the module as electrical signals, and wherein the optical input/output connectors of the module bypass circuit boards of the first and second mass storage modules are connected by a fiber optic transmission medium such that signals are communicated between the modules in the form of light; 004880335 3b said first controller connecting the at least one CPU of the first controller with each said disk drive through its associated drive bypass circuit board and through the module bypass Scircuit board such that a loop is formed between the output and input of the first controller with Seach disk drive bypass circuit board and the module bypass circuit board in said loop and 0 5 completing said loop whether or not a disk drive is plugged into the disk drive connector; and 00said second controller connecting the at least one CPU of the second controller with each said disk drive through its associated drive bypass circuit board and through the module bypass Scircuit board; and Sat least one of said first and second servers being operative to establish direct communication between the first and second controllers, and said first and second controllers being operative to maintain direct communication between the first and second controllers independent of said at least one CPU of said first server and said at least one CPU of said second server.

Certain information server configurations in accordance with embodiments of the invention include one or more computers, each computer connected to communicate through a Fibre Channel controller with a mass storage array comprising a plurality of bypass circuit boards, at least some of which are connected to an information storage device. In one embodiment, the controller provides a dual loop communications channel comprising two complete communication paths to each of bypass circuit boards and associated storage devices.

In another embodiment, the controller provides a single loop which traverses the bypass circuit boards and any associated storage devices twice. In a one configuration with two or more computers, the Fibre Channel controller connected to each computer communicates with the mass storage array through a Fibre Channel controller bypass card.

In a preferred embodiment the computer is preferably a suitable conventional NlJfli board computer. The controller preferably is a conventional arbitrated dual channel Fibre Channel system through which the computer communicates With each of tesoae; device bypass circuit boards and the module bypass circuit board. The bypass circuit boards may be any suitable circuits which form a continuous loop for the 00 Fiber Channel controller regardless of whether a disk drive is plugged into the drive bypass circuit board. The ioop continues through other modules when they are connected to the module bypass circuit thereby readily permitting expansion while maintaining a unitary information processing and mass storage system.

0 10 In accordance with another embodiment of the present invention, a high speed information processing and mass storage system includes two modules, each including a plurality of disk drives in a hot-swappable, disk drive array. Each disk drive array is connected to a module bypass circuit which includes an optical input/output connector, preferably an optoelectronic transceiver. The optical input/output connectors of the modules are connected by a fiber optic transmission medium such that signals are communicated between the modules in the form of light. With this configuration, modules may be added to increase storage capacity without interrupting the operation of each other and without serious signal degradation.

In accordance with yet another embodiment of the present invention, a high speed information processing and/or mass storage system with disk drives for information storage includes at least one module with a plurality of drive bypass circuit boards, each including a drive bypass circuit board connector. At least one opening is provided between connectors to permit the flow of air between the connectors. Each drive bypass circuit board is a relatively flat circuit board with connectors on different edges of the board, wherein one of the connectors is the connector which receives the disk drive and the other connector connects to said drive bypass circuit board connector.

The connectors, bypass circuit boards and drives are arranged such that when they are connected there is a path for air flow from outside the module alongside each bypass circuit board and its associated disk drive for cooling purposes without any backplane obstruction. Where the mass storage system is housed in an enclosure, at least one fan is mounted to force air from outside said enclosure through the spaces between said 004880335 bypass boards and drives, preferably through the openings between the drive bypass circuit r board connectors.

SIt will be appreciated that embodiments of the present invention provide a novel high 0 speed information processing and/or mass storage system particularly suitable for information 5 and application servers. The system is scalable, fault tolerant, and reliable both because of its 0 00 novel system architecture and its physical layout. Other features and advantages of the invention will become apparent from the following detailed description of exemplary and preferred N embodiments when read in conjunction with the drawings which illustrate, by way of example, the principles of the invention.

BRIEF DESCRIPTION OF THE DRAWINGS Figure 1 is a functional block diagram generally illustrating an information or application server system according to the present invention using a high speed mass storage system of the invention; Figures 2A, 2B and 2C are diagrammatic representations illustrating various system configurations for different applications wherein different ratios of processing and storage are required; Figure 3 is a functional block diagram illustrating a mass storage module of Figure 1 in greater detail; Figure 4A is a functional block diagram illustrating the dual loop communication path from the Fibre Channel controller of Figure 3 connected so as to provide one logical Fibre Channel communication path traversing each of the storage device bypass circuit boards twice; Figure 4B is a functional block diagram illustrating the dual loop communication path from the Fibre Channel controller of Figure 3 connected so as to provide two logical Fibre Channel communication paths traversing each of the storage device bypass circuit boards once, thereby being independently available to communicate with each of the storage devices; 004880335 Figure 5 is a functional block diagram illustrating an information or application server system according io the present invention wherein two or more servers are connected to a mass P storage array which may include one or more mass storage modules; 00 t",D t",l 0", 0 Figure 6 is a functional block diagram illustrating an embodiment of a web Cl e:Vic;i application configured in accordance with the architecture of the present CD invention; C Figure 7 is a functional block diagram illustrating an embodiment of a basic video streaming application configured in accordance with the architecture of the present oO invention; N Figure 8 is a functional block diagram illustrating another embodiment of a video 0streaming application configured in accordance with the architecture of the present invention and including a duplicated index server; O 10 Figure 9 is a functional block diagram illustrating another embodiment of a video streaming application configured in accordance with the architecture of the present invention and including a distributed index server; Figure 10 is a functional block diagram illustrating one embodiment of a storage device bypass circuit board according to the present invention; Figure 11 is a functional block diagram illustrating one embodiment of a chassis bypass circuit board according to the present invention; Figures 12 A and 12 B are diagrammatic representations of the physical layout of a drive bypass circuit board and disk drive illustrating the preferred connector arrangement according to the present invention; and Figures 13 A and 13 B are diagrammatic representations of a connector arrangement for connecting a single board computer (SBC) to various input/out devices according to the present invention.

DETAILED DESCRIPTION OF PREFERRED EMBODIMENTS One embodiment of an information or application server system in accordance with the present invention using the high speed information processing and mass storage system of the invention is illustrated in Figure 1. Referring to Figure 1, the server, designated generallybythe reference numeral 100, includes a computer 102, a controller 104, preferably a Fibre Channel controller, and a communications interface or access card 106. The server 100 communicates via the Fibre Channel controller 104 with a mass storage array 200 which includes one or more mass storage modules 202A O 202n. As illustrated, the computer 102 may also communicate with a suitable diagnostic c computer 110 as is described in the aforementioned Dellacona applications.

DThe computer 102 preferably comprises single board high speed computer runming a computer industry standard operating system software program such as, for example, WindowsNT available from Microsoft Corporation. An operating system like 0) Windows NT may be stripped down to remove those elements of the program which are Snot needed, if desired to preserve memory or to increase operating speed. Suitable Sconventional drivers of the type used for similar applications may be provided as r necessary to support the particular architecture being implemented.

The computer may include a display such as a touch screen display, and various storage and peripheral devices (not shown) as required. The single board computer can include any of a wide number of suitable devices including, but not limited to, the Compact PCI CPU Board with Pentium Processor, Model No. ZT 5510, available from Ziatech Corporation. Modifications to enhance performance of the ZT 5510 can include an onboard 40 MB flash memory card for permanent storage of the non-reconfigurable portions of the Windows NT operating system software and an onboard, removable, PCMCIA 40 Mb flash memory card, "D2 FlashDisk" available from Sandisk SCorporation for read/writeable storage of the reconfigurable portions of the Windows NT software.

The Fibre Channel controller 104 may be any suitable design according to the Fibre Channel Consortium created as a separate board or incorporated into the single board computer design. The communications interface or access card 106 may be any suitable device made in accordance with well known T-l communications architecture, and/or architecture adapted for compatibility with other network and telecommunications architectures, protocols and topologies including, but not limited to, T-3, DS-3, OC-3 C, OC-12C, OC-192C, FDDI, SONET, SCSI, TCP/IP, HiPPI and ATM. In addition, the computers 102 and 110 may be networked together and with other computers through appropriate ethemet cards or other suitable networking techniques. The respective manufacturer, Fibre Channel Consortium and 120 Special Interest Group reference design data sheets and materials describing the detailed operating capabilities and specifications of each of the foregoing components are hereby incorporated by reference in their entirety. Also, further information concerning 0 .possible subsystems and connection protocols for the server are described in the

O

C reereenced Dellacona applications.

Figures 2A, 2B and 2C illustrate generally three types of system configurations which can be addressed, in accordance-with the architecture of the present invention.

Each Figure is diagrammaticallyillustrative of a chassis 300 with apower supply section 302 and a diagnostics section 304. In addition, the chassis includes a processing section ND 3 06 such one or more of the servers 100 of Figure 1, and /or a storage section 308 such Sas one or more of the mass storage modules 202 of Figure 1. The size of the processing and storage sections can be readily configured for a particular application.

0 10 For example, Figure 2A illustrates an arrangement for a high volume server such as an application server, a movie server for video streaming to multiple users) or communications in conjunction with a carrier-class switch. It can be seen that the processing section 306 includes ten slots, each preferably representing a single board computer. In contrast, the storage section 308 has only five storage slots, each representing a high speed, high capacity storage device such as a disk drive. Thus, the processing function is stressed over the storage function. On the other hand, Figure 2B illustrates an arrangement which might be suitable for a webserver or web-hosting.

Here, the processing section 306 comprises two slots of the chassis whereas the storage section 308 comprises fifteen slots. Figure 2C illustrates a chassis arrangement suitable primarily for storage expansion. Here, there is no processing section and the chassis is devoted to storage.

Figure 3 illustrates an embodiment of the mass storage array 200 of Figure I in greater detail as it could be configured with a single computer server 100 and one or more mass storage modules 202. Referring to Figure 2, the mass storage array 200 includes one or more modules 202A, 202B 202n, each of which is preferably identical architecturally, although they may contain different numbers of storage devices and different types of storage devices. Each module 202 includes bypass circuit boards 210A,210B 21 On and at least some storage devices 212A, 212B 212n connected to the boards 210. In addition, a each storage device is preferably provided with an associated read/write switch 214A, 214B 214n, respectively.

Each module 202 has a module or chassis bypass circuit board 216 in the communication path of the Fibre Channel controller 104. The chassis bypass circuit board 216 of module 202A is provided with an optical input/ output connector 218 for outputting electrical signals from the module 202A as light signals and for inpuitting1 Light signals into the module 202A as electrical signals. Likewise, the modules 202B N- 202n have chassis bypass. circuit boards 216 with associated input/output connectors 218 (not shown). As illustrated, the input/output connector 218 of the chassis bypass 00 circuit board ofrnodule 202A is adaptedto be connected via a light transmission medium such as optical fibers 220 to the optical input/output connect-or 2 18 (not shown) of the chassis bypass circuit board of module 202B.

As was previously noted, the controller 104 preferably is a conventional Fibre Channel Controller (FCC) which operates on a Fibre Channel protocol, and preferably is an arbitrated dual channel Fiber Channel Controller. The controller 104 provides a dual channel communication path within each module 202 between the computer 102 and each of the operable storage devices 212. As is described hereinafter in greater detail, the bypass circuit boards 210 ensure that the communication path is complete even if a storage device 212 is inoperable is not operable at or above some minimum level as is hereinafter described in greater detail) or has been removed from the connector on the bypass circuit board.

In that regard, each of the storage devices 212 is preferably a high speed, high capacity, conventionally available disk drive which is removably connected to its associated bypass circuit board 210. Preferably, the disk drive plugs into a connector on the bypass circuit board 210 so that it can be readily removed and replaced or so that drives may be added to empty bypass circuit board positions, as needed to expand the storage capacity of the module. Each bypass circuit board 210 includes circuits which connect the controller 104 to the disk drive 212 when the disk drive is plugged in and is conveying to the bypass circuit board that it is operable at a certain minimum level.

On the other hand, the bypass circuit board 210 connects the controller 104 directly to the next bypass circuit board, bypassing the disk drive 212, when the disk drive is not plugged in or is not operating at or above the minimum level. Any suitable, conventional disk drive of the type that runs self-diagnostics and provides a fault/no fault output signal may be used for this purpose.

With futher reference to Figure 3, by way of example, the bypass circuit board 210OA directs communications to the associated storage device 21 2A and then to the next O bypass circuit board 210B when an operable storage device 212A is connected to the "ypass circuit board 2i0A. When the storage device 212A is inoperable, i.e. not operating at some minimum satisfactory level, or when there is no storage device CI -connected to the bypass circuit board 210A, if the storage device is removed for replacement, the dual channel communication path proceeds through the bypass circuit 00 board 210A without interruption, bypasses the storage device connector. While one embodiment of a bypass circuit board to accomplish the foregoing connection and 0 bypass functions is described hereinafter in greater detail, it will be appreciated by one skilled in the art that these functions can be accomplished in any suitable conventional manner by electronic switching circuits controlled by the fault/no fault signal from the storage device.

With continued reference to Figure 3, the module or chassis bypass circuit board 216 provides functions similar to the storage or drive bypass circuit boards 2 10 in the sense that they either route communications back to the controller directly if there is no additional module 202B connected to the module 202A or they route communications to the next module 202B if one is connected and signals are being received. In the case of a single module 202, therefore, there is a dual channel communications loop entirely within the module by which the computer 102 communicates with each of the disk drives 212. On the other hand, byvirtue of the chassis bypass circuit board 216, the dual channel communications loop extends to each of the disk drives 212 in the next module when one is connected. Specifically, if there is an additional module 202B connected to module 202A as illustrated, then communications from the computer 102 of module 202A are directed by the module bypass circuit board 216 to the next module 202B via connectors 218 and optical fibers 220 so that module 202B is within and traversed by the dual Fibre Channel communication loop.

It will be appreciated that with the above described architecture, individual mass storage device modules 202 of the mass storage array 200 may be expanded internally by adding disk drives or other suitable storage devices, and bad disk drives may be replaced without affecting the operation of the rest of the module or the system it is used in. This provides an extremely versatile hot swappable, hot expandable, mass storage device array. In addition, when the demands of the system exceed the capacity of a O single module, an additional module may be added, again without interrupting the c operation of the rest of the array or its system.

.As was explained above, the Fibre Channel controller provides a dual path through each module of the mass storage array 200. In accordance with the present invention, the system can be configured so that the dual path is a single path which OC traverses the mass storage array twice or two independent paths as is illustrated in Figures 4A and 4B.

Referring now to Figures 4A and 4B, the Fibre Channel controller (FCC) 104 includes two output paths A and B. Each path traverses the mass storage module 202 as illustrated, communicating with each of the present and operable storage devices 212 and returning to the FCC. In the Figure 4A embodiment, the A path returning to the FCC is looped back to the B output path. Accordingly, a single continuous path traverses the mass storage module twice. In the Figure 4B embodiment, the A return path is not looped back to the B output path. Accordingly, two paths traverse the mass storage module and can be used independently. This latter embodiment provides a second path in the event that one path fails.

It will be appreciated by one skilled in the art that the Figure 4A and 4B embodiments provide redundancy and fault tolerance. The Figure 4A embodiment is somewhat simpler to implement because only one set of chips is necessary to provide the single Fibre Channel capability required for the single loop. Still, if one of the loops is broken or encounters some other fault, that loop can be bypassed within the controller, and the other loop is still available. Similarly, while the Figure 4B embodiment may be more complex and expensive to implement, it provides two independently addressable loops for fault tolerance and redundancy, but also provides significantly greater communications bandwidth.

Figure 5 illustrates a further configuration which is made possible by the system architecture of the present invention. In the Figure 5 embodiment, two servers are connected to a mass storage array, a first module of which is illustrated. The servers 100A and 100B are connected to Fibre Channel bypass boards 222A and 222B, respectively. The operation of the computers 102A and 102B may be sensed by the Fibre Channel bypass boards, for example by sensing signal flow as with the chassis bypass board, so that the computers 102A and 102B can operate together or, in the case of a fault, separately. As was previously mentioned, each server may have flash N rnczucry; and irithe embo'dimentof Figure 5, each computer 102A and 102B has its own flash memory with its operating system, stored therei. In this fashion, the computers can boot independently from its associated flash memory rather than from the shared memory or other disk arrangement.

00 It will be appreciated that the system configuration illustrated in Figure provides two servers and thus increased processing power. In addition, one server provides backup to the other in the event of failure, thereby providing increased fault N ~tolerance and reliability. In addition, the architecture ofFigure 5 permits the two servers S 10 to communicate with each other at Fibre Channel speeds, higher than could, be achieved with IlOOBaseT LAN, using IP protocol.

It can be seen that system architecture according to the present invention lends itself to a wide variety of configurations to accommodate a variety of applications.

Figure 6 illustrates one configuration wherein two servers A and B are connected in a suitable local area network (LAIN) configuration with access to the internet. Server A has its own mass storage array 200 as does server B. It will be appreciated that several servers and storage arrays can be networked in this fashion to provide a very powerful information or application server system particularly suitable for web server applications. Inr addition, this configuration, permits fulfl duplication of computer/mass storage systems rather than sharing a mass storage system as with the embodiment of Figure 5. Moreover, while the Figure 6 embodiment illustrates communication over a LAN which typically may be a lOOBaseT LAN, it will be appreciated that this may be a Fibre Channel LAN if rates in the gigabit range are desired. These features and advantages may be the ideal choice for critical processing applications such as web servers.

Figure 7 illustrates another configuration particularly useful for video streaming applications. In this embodiment the diagnostics computer 110 also has indexing funrctions for controlling access to content servers 1, 2 and 3. The content servers 100 provide access to video stored in their associated storage arrays 200. Alternatives, also particularly suitable for video streaming applications, are shown in Figures 8 and 9. In the Figure 8 embodiment duplicated index servers separate from the diagnostics computer are provided with their own storage arrays. In the illustrated embodiment there are two index servers I and 2 and there are three content servers 1, 2 and 3, each with an associated storage array. In theFigure 9 embodiment, the functions of the index Sservers are distributed among the content servers so that there are multiple index/content C1 servers 1-5. One skilled in the art will appreciate that the Figure 8 approach uses an architecture where the index server or servers coordinate content streaming from the 00 content servers whereas in Figure 9 the index and content server functionality is distributed across all servers, providing extensive scalability.

SFigures 10 and 11 are functional block diagrams which generally illustrate the storage and chassis bypass circuit boards 210 and 216, respectively, in greater detail.

Referring to Figure 10, the storage bypass circuit board 210 includes a bypass board backplane connector 230 arranged to plug into a connector on the backplane generally indicated at 232. The A and B signal paths coming from a previous bypass circuit or directly from the Fibre Channel controller are connected through the backplane to the bypass board via the backplane connector 230. Similarly, the A and B signal paths emerge from the bypass board 210 via the bypass board backplane connector 230.

The A signal path from the backplane connector 230 is connected to a suitable conventional electronic switch 234. The B signal path from the backplane connector 230 likewise is connected to an electronic switch 236. The A and B signal paths from electronic switches 234 and 236 are connected to a bypass board storage card or drive connector 238 where they are routed to the storage device a disk drive) 212.

The return A signal path from the bypass board drive connector 238 is connected to the switch 234, and the return B signal path from the connector 238 is connected to switch 236. A fault signal produced by the storage device to indicate its presence and its level of operability as was described above is applied to each of the electronic switches 234 and 236 to control the switching thereof. The A and B return paths from the switches 234 and 236 are connected to the bypass board backplane connector 230 where they are routed through the backplane 232 to the next bypass circuit board or to the Fibre Channel controller.

In operation, the A signal path enters the bypass circuit board and is connected to the switch 234. If the fault signal is not present there is no fault and the signal is in a low or negative signal state) indicating that the storage device is not present or is inoperable, the switch 234 returns the A signal path to the bypass board backplane b connector 230 thus bypassing the storage device 212. The B signal path similarly is C l:ope biack'to the backplane connector 230 by the electronic switch 236 if the fault signal is low. On the other hand, the A and B signal paths are routed through the switches 234 and 236 to the storage device and then back through the switches when the 0 5 fault signal is high or positive indicating the storage device is present and operable.

00 Referring now to Figure 11, the chassis bypass circuit board is essentially the ,0 same as the storage bypass circuit board except the selection made by the electronic 0 switches 234 and 236 is between acting as a bypass or connecting the A and B to the connector 238. In this regard, the 1/0 connector is preferably a conventional optical 0 10 fiber transceiver for use in bi-directional communication applications over multimode optical fiber, particularly in multimode or single mode Fibre Channel applications. For example, the transceiver may be a model MLC-25-6-X-T optical fibre channel small factor (SFF) transceiver available from Methode Electronics, Inc. of Chicago, Illinois.

Such transceivers include a light transmitter and receiver as well as a standard receptacle for receiving an industry standard optical fiber connector. In addition, the transceiver provides a signal detect output (the fault signal in Figure 11) which indicates whether or not the transceiver is receiving a light signal. If it is not, the fault signal causes the electronic switches 234A and 236A to loop the A and B paths back to the chassis bypass board connector 230A. If, on the other hand, the fault signal indicates that a light signal is being received by the transceiver 218, the electrical signals on the A and B paths are passed through the switches 234A and 236A to the transceiver 218 where they are converted to light signals and transmitted over the optical fibers forming the A and B paths to the next chassis. Similarly, light signals returned from the next chassis on the A and B paths are converted back to electrical signals by the transceiver 218. and returned along the A and B paths through the switches 234A and 236A to the connector 230A and onto the next bypass circuit or the Fibre Channel controller.

Typically, each of the modules 202 of a mass storage device array such as the one shown in Figure 3 is enclosed in a housing or, if mounted in a rack, may be surrounded by other structures and/or other circuit boards. The server components such as the computers 102 and 110, the touch screen display 112, the Fibre Channel controller, and other system components such as communication access cards, ethernet cards, LAN components and the like may also be mounted within the same housing or on the same rack. Heating may therefore be a problem, particularly where the storage devices used in the modules are disk drives driven by motors. Accordingly, the d9 preferred embodiment of the present invention as illustrated in Figures 12A and 12B C .includes a physical structure and arrangement of the various bypass circuit boards and storage devices that accommodates the circulation of cooling air throughout the module 00 without minimal obstruction.

In accordance with one aspect of this physical structure of the invention, each of Sthe drive bypass circuit boards 210 is a relatively thin circuit board. The circuit board is, however, unlike typical circuit boards built to receive a disk drive or other mass storage device. In such conventional circuit boards, the connector or plug which receives the plug-in disk drive typically is positioned so that the disk drive extends perpendicular to the plane of the board. For example, it is usual to have the bypass circuits and/or the communications paths between them on a backplane or midplane circuit board which extends across the module in a fashion similar to a computer motherboard which extends across the computer chassis and has connectors to receive various plug-in cards or boards. That arrangement creates an obstruction which makes it more difficult to effectively cool heat producing storage devices such as disk drives.

As, is illustrated in Figures 12A and 12B, the present invention does not use either a circuit board midplane or a backplane structure. Instead, the connector on the drive bypass board which receives the disk drive the connector 238 in Figure is at the edge of the board so that the drive, when plugged in, extends parallel to the circuit board. In addition, the drive bypass circuit board itself is not plugged into a backplane circuit board. Rather, a connector is provided on the edge of the board, preferably opposite the disk drive connector as illustrated at 239 in Figure 10, and that connector plugs into an individual connector which is mounted on a frame or other structural member of the chassis and which is wired or otherwise connected to similar connectors for the other disk bypass circuit boards.

Figures 12A shows a side view of the disk drive 212 plugged into the drive bypass circuit board 210 via connector 23 8, with the drive bypass circuit board plugged into connector 240 suitably mounted on structural members 242 of the chassis or rack containing the mass storage module and/or server and its associated components. Figure 12 B is an end view illustrating several bypass circuit boards 21 plugged into the 0 connectors 240 which are in turn connected by screws or other suitable means to the N

C

structural members 242. Since there is no backplane which would normally make the connections between adjacent components, electrical or light connections generally .indicated at 246 are suitably provided between the connectors 240 to provide the communications required, as illustrated, for example, in Figure 3 and OC It can be seen that between each connector 240 there is a space 244 through Swhich air can readily be drawn or forced by a fan or other air circulation means as is O necessary. Even if the structural members are part of a housing that surrounds the components, screening or other suitable openings can be provided so that the areas 244 permit sufficient air flow. It can also be seen that because the disk drive is plugged into the bypass circuit board so that their planes are parallel and not perpendicular, there is no obstruction of air flow.

It will also be appreciated that this arrangement is particularly suited for field service of the unit and is readily upgradeable. Connectors can be readily replaced in the field without the need to change a complete backplane or midplane board, and in some instances repairs of this sort can be carried out with little or no down time. In addition, the illustrated connection arrangement permits expansion without the limitations encountered when using a backplane or midplane with a fixed number of expansion slots and without the other physical and electrical limitations encountered with backplanes or midplanes.

Similar physical arrangements may be used to connect computers to their associated components to create the desired server configuration. For example, as illustrated in Figure 13A, a single board computer of the type previously described in connection with the description of the server 100 may connect to the communications card 106 through a mini compactPCI (CPCI) backplane. In this case, the single board computer SBC may incorporate the Fibre Channel controller FCC. Likewise, in Figure 13B, the SBC may connect to an input/output unit 1/0 at the left rear connector of the CPCIbackplane. In this case, the 1/0 may include the FCC functions. Again, it will be appreciated that the foregoing advantages are achieved with this sort of simple backplane or midplane structure which does not extend across the cabinet or rack.

The above-described exemplary embodiments are intended to be illustrative in all, respects, rather than restrictive, ofthe present invention., Thus the present invention 004880335 17 is capable of many variations in detailed implementation, that can be derived from the C<1 description contained herein by a person skilled in the art. All such variations and modifications are considered to be within the scope and spirit of the present invention as defined by the following claims.

It will be understood that the term "comprises" (and its grammatical variants) as used in ,0 the specification is equivalent to the term "includes" and should not be taken as excluding the existence of other elements or features.

0",

(N-

1^,

Claims (42)

1. A method for transferring data in a network server system, the network server system being connected to a network for providing subscribers on a network with access to data from the network server system, comprising: 00 O 5 providing at least one mass storage device; providing a mass storage device controller, said mass storage device controller being connected to said at least one mass storage device for controlling input and Soutput of said at least one mass storage device; providing a communications interface, said communications interface providing a connection to the network for receiving input signals from the network and for outputting signals from the network server system to the network, said mass storage device controller being connected to said communications interface for receiving input signals from the network and for outputting signals from said at least one mass storage device to the network; establishing direct communication between said mass storage device controller and said communications interface; and maintaining the direct communication between said mass storage device controller and said communications interface to provide the subscribers on the network with access to data from the network server system.

2. The method of Claim 1, wherein said mass storage device controller operates with a Fibre Channel protocol.

3. The method of Claim 1, wherein said mass storage device controller comprises an arbitrated loop dual channel Fibre Channel controller.

4. The method of Claim 1, wherein said mass storage device controller is connected to an optical input/output connector which is connected to said at least one mass storage device, the optical input/output connector outputting electrical signals as output 0-4928114 19 light signals and inputting light signals to said mass storage device controller as input Nelectrical signals. A network server system for transferring data, the network server system being 0connected to a network for providing subscribers on the network with access to data from the network server system, comprising: 00 Sat least one mass storage device; 0 a mass storage device controller, said mass storage device controller being c connected to said at least one mass storage device for controlling input and output of said at least one mass storage device; a communications interface, said communications interface providing a connection to the network for receiving input signals from the network and for outputting signals from the network server system to the network, said mass storage device controller being connected to said communications interface for receiving input signals from the network and for outputting signals from said at least one mass storage device to the network; means for establishing direct communication between said mass storage device controller and said communications interface; and means for maintaining the direct communication between said mass storage device controller and said communications interface to provide the subscribers on the network with access to data from the network server system.

6. The network server system of Claim 5, wherein said mass storage device controller operates with a Fibre Channel protocol.

7. The network server system of Claim 5, wherein said mass storage device controller comprises an arbitrated loop dual channel Fibre Channel controller.

8. The network server system of Claim 5, wherein said mass storage device controller is connected to an optical input/output connector which is connected to said at '01928114 O least one mass storage device, the optical input/output connector outputting electrical N signals as output light signals and inputting light signals to said mass storage device Scontroller as input electrical signals.

9. A network server system, comprising: 00 a central processing unit; N a mass storage device; Sa first controller, said first controller being connected to said mass storage device for controlling input and output of said mass storage device; a communications interface, said communications interface providing a connection to the network for receiving input signals from the network and for outputting signals from the network server system to the network; and a second controller communicatively coupled to the central processing unit and the first controller, the second controller communicatively coupled to said communications interface to communicate with a network, the central processing unit being operative to establish direct communication between the first and second controllers, and said first and second controllers being operative to maintain the direct communication independent of the central processing unit. The network server system of Claim 9, wherein the first controller communicates with the mass storage device over a high speed optical network.

11. A network server system connected to a network for providing subscribers on the network with access to data from the network server system, comprising: a central processing unit; a network communications interface; a mass storage device; and 004928114 21 a storage device controller communicatively coupled to the network communications interface and to the mass storage device to control communications between network communications interface and the mass storage device, said central processing unit being operative to establish direct communications between the mass storage device and the network communications interface via the storage device controller to provide the subscribers on the network with access to data from the network server system.

12. A method for transferring data in a network server system, the network server system being connected to a network for providing subscribers on a network with access to data from the network server system, comprising: providing at least one mass storage device; providing first and second controllers, at least one of the first and second controllers being connected to the network for receiving input signals from the network and for outputting signals from the network server system to the network, and at least one of the first and second controllers being connected to said at least one mass storage device for controlling input and output of said at least one mass storage device; providing at least one central processing unit connected to the first and second controllers, said at least one central processing unit establishing direct communication between the first controller and the second controller; and maintaining the direct communication between the first and second controllers independently of said at least one central processing unit, freeing said at least one central processing unit.

13. The method of Claim 12, wherein the first and second controllers operate with a Fibre Channel protocol.

14. The method of Claim 12, wherein the first and second controllers comprise arbitrated loop dual channel Fibre Channel controllers. C04928114 22 The method of Claim 12, wherein said at least one of the first and second controllers is connected to an optical input/output connector which is connected to the at 0 least one mass storage device, the optical input/output connector outputting electrical Ssignals as output light signals and inputting light signals to the first controller as input electrical signals. 00

16. A method for transferring data in a network server system, the network server system being connected to a network for providing subscribers on a network with access to data from the network server system, comprising: C providing a high speed mass storage system; providing first and second Fibre Channel controllers, at least one of the first and second controllers being connected to the network for receiving input signals from the network and for outputting signals from the network server system to the network, and at least one of the first and second controllers being connected to the high speed mass storage system for controlling input and output from the mass storage system; providing at least one central processing unit connected to the first and second Fibre Channel controllers, said at least one central processing unit establishing direct communication between the first Fibre Channel controller and the second Fibre Channel controller; and maintaining the direct communication between the first and second controllers independently of said at least one central processing unit, freeing said at least one central processing unit.

17. The method of Claim 16, wherein the first and second Fibre Channel controllers comprise arbitrated loop dual channel Fibre Channel controllers.

18. The method of Claim 16, wherein said at least one of the first and second Fibre Channel controllers is connected to an optical input/output connector which is connected to the high speed mass storage system, the optical input/output connector 004928114 23 outputting electrical signals as output light signals and inputting light signals to the first controller as input electrical signals.

19. A method for transferring data in a network server system, the network server system being connected to a network for providing subscribers on a network with access to data from the network server system, the method comprising: 00 Oproviding a mass storage system which is readily expandable to increase its N storage capacity while the system is in operation, said mass storage system including at Sleast one mass storage module with a plurality of plug-in storage devices for storing Sinformation; providing at least one central processing unit; providing a plurality of storage device bypass circuit boards associated with each of said storage devices, respectively, each storage device being plugged into a connector on the storage device bypass circuit board; providing a module bypass circuit board including an optical input/ output connector for outputting electrical signals from said at least one mass storage module as light signals and for inputting light signals into said at least one mass storage module as electrical signals; and providing at least one controller providing a communication path between said at least one central processing unit with said plurality of storage devices through said storage device bypass circuit boards, respectively, and through the module bypass circuit board. The method of claim 19 wherein each storage device bypass circuit board includes a circuit which completes the connection of the CPU with the other storage device bypass circuits and their associated storage devices whether or not the storage device is present. 004928114 24 21 The method of claim 19 wherein said at least one mass storage module Ncomprises a first mass storage module and at least one additional mass storage module, and the module bypass circuit board connects to said at least one additional mass storage module by outputting electrical signals from said first mass storage module to said at least one additional mass storage module via the optical input/output connector when light signals are received from said at least one additional mass IN storage module by said optical input/output connector.

22. The method of claim 19 wherein said at least one mass storage module comprises first and second mass storage modules each including one said module bypass circuit board including one said optical input/output connector, and wherein the optical input/output connectors of the first and second mass storage modules are connected by a fiber optic transmission medium such that signals are communicated between the first and second mass storage modules in the form of light.

23. The method of claim 22 wherein the module bypass circuit board of the first mass storage module connects to the second mass storage module by outputting electrical signals from the first mass storage module to the second mass storage module via the optical input/output connectors when light signals are received from the second mass storage module by said optical input/output connector of said first mass storage module.

24. The method of claim 19 wherein the controller operates with a Fibre Channel protocol. The method of claim 19 wherein the controller is an arbitrated loop dual channel Fibre Channel controller.

26. The method of claim 19 wherein each storage device is a disk drive and wherein each storage device bypass circuit board comprises a disk drive bypass circuit board including a circuit which completes the connection of the CPU with the other drive bypass circuits and their associated disk drives whether or not the disk drive is present.

27. The method of claim 26 wherein said at least one mass storage module comprises a first mass storage module and at least one additional mass storage 001928114 0 module, and the module bypass circuit board connects to said at least one additional N mass storage module by outputting electrical signals from said first mass storage Smodule to said at least one additional mass storage module via the optical input/output Nconnector when light signals are received from said at least one additional mass storage module by said optical input/output connector. 00 S28. The method of claim 25 including first and second mass storage modules each Oincluding one said module bypass circuit board including one said optical input/output N connector, wherein the optical input/output connectors of the first and second mass Sstorage modules are connected by a fiber optic transmission medium such that signals are communicated between the first and second mass storage modules in the form of light.

29. The method of claim 28 wherein the module bypass circuit board of the first mass storage module connects to the second mass storage module by outputting electrical signals from the first mass storage module to the second mass storage module via the optical input/output connectors when light signals are received from the second mass storage module by said optical input/output connector of said first mass storage module. A network server system, comprising: a central processing unit; a first controller communicatively coupled to the central processing unit; a mass storage device communicatively coupled to the first controller, the first controller configured to control communications to and from the mass storage device; and a second controller communicatively coupled to the central processing unit and the first controller, the second controller configured to communicate with a network, the central processing unit being operative to establish direct communication between the first and second controllers, and said first and second controllers being operative to maintain the direct communication independent of the central processing unit. C04928114 26 31 The network server system of Claim 30, wherein the first controller communicates Nwith the mass storage device over a high speed optical network.

32. A server system, comprising: a central processing unit; 00 IND a first controller communicatively coupled to the central processing unit and Sconfigured to control communications to and from at least one mass storage device Sover an optical communication path; and a second controller communicatively coupled to the central processing unit and the first controller, the second controller configured to communicate with a network, the central processing unit being operative to establish direct communication between the first and second controllers, and said first and second controllers being operative to maintain the direct communication independent of the central processing unit.

33. In a network server system, the improvement in the network server system comprising: a first controller communicatively coupled to the network server system; a mass storage device communicatively coupled to the first controller, the first controller configured to control communications to and from the mass storage device; and a second controller communicatively coupled to the network server system and the first controller, the second controller configured to communicate with a network, the network server system being operative to establish direct communication between the first and second controllers, and said first and second controllers being operative to maintain the direct communication with each other once the direct communication is established.

34. A network server system, comprising: 004928114 27 r"- O a network communications interface; d a mass storage device; and (Ni a storage device controller communicatively coupled to the network communications interface and to the mass storage device to control communications 00 O 5 between the network communications interface and the mass storage device, said 0network communications interface being operative to establish direct communications C between the mass storage device and the network communications interface via the storage device controller. The network server system of Claim 34, further comprising: a central processing unit communicatively coupled to the network communications interface to receive data requests from the network communications interface.

36. The network server system of Claim 35, wherein the central processing unit is bypassed by establishing direct communications between the mass storage device and the network communications interface via the storage controller.

37. A method for transferring data in a network server system, the network server system being connected to a network for providing subscribers on a network with access to data from the network server system, the method comprising: providing a network communications interface; providing a mass storage device; providing a storage device controller communicatively coupled to the network communications interface and to the mass storage device to control communications between network communications interface and the mass storage device; and establishing direct communications between the mass storage device and the network communications interface via the storage device controller. 004928114 28

38. A network server system, comprising: Sa network communications interface; Sa mass storage device; o00 ND a storage device controller communicatively coupled to the network communications interface and to the mass storage device to control communications C between network communications interface and the mass storage device; and Smeans for establishing direct communications between the mass storage device and the network communications interface via the storage device controller.

39. A mass storage server, comprising: a plurality of interface cards, each interface card configured to automatically detect whether a storage device is coupled to the interface card; and a mid-plane connector board having two opposing sides and a plurality of sockets for connecting the interface cards on each opposing side of the mid-plane connector board.

40. The mass storage server of Claim 39, wherein the plurality of interface cards are bypass cards that can be sequentially connected together to interconnect a plurality of storage devices.

41. A mass storage server, comprising: a plurality of hot-swappable storage devices interconnected in a loop; and a mid-plane connector board having two opposing sides and a plurality of sockets for connecting the hot-swappable storage devices on each opposing side of the mid-plane connector board. 004928114 29

42. The mass storage server of Claim 41, wherein each of said hot-swappable storage devices is coupled to the mid-plane connector board via a bypass card.

43. An information server system having a scalable, modular, fault tolerant, hot swappable architecture of a plurality of components for interfacing with a computer 5 network, comprising: 0O a central processing unit; means for interfacing with a computer network connected to the central N processing unit; a mass storage subsystem connected to the central processing unit; and a mid-plane connector board having two opposing sides and means for connecting the interface cards for the components on each said opposing side of the mid-plane connector board.

44. Apparatus for increasing the throughput rates of a user computer having a communications interface via a network with a host server system, the user computer communications interface including a modem of the type utilizing a database hash table for decryption of encrypted data received from the host server system, the apparatus comprising: means for installing a supplementary database hash table in the user computer to replace the function of the hash table in the modem; means for accessing the supplementary hash table installed in the computer for decryption of encrypted data received from the host server system; and means for synchronizing the modem with the transmission speed of the host server system by gradually increasing the setting of the throughput rate of the modem along with that of data transmission from the host server system.

45. A system for increasing the throughput rates of a user computer, comprising: 004928114 0 a user computer having a communications interface, the communications Sinterface including a modem of the type utilizing a database hash table for decryption of 0 encrypted data received; Sa host server system; and 00 O 5 a network communicatively coupling the host server system and the user acomputer, wherein a supplementary database hash table is installed in the user N computer to replace the function of the hash table in the modem, the supplementary 0 hash table installed in the computer is accessed by the user computer to decrypt N encrypted data received from the host server system, and the modem is synchronized with the transmission speed of the host server system by gradually increasing the setting of the throughput rate of the modem along with that of data transmission from the host server system.

46. A high speed mass storage system which is readily expandable to increase its storage capacity while the system is in operation, comprising: first and second mass storage modules, each mass storage module having at least one hot-swappable storage device; a module bypass circuit board including an optical input/output connector for outputting electrical signals from the module as light signals and for inputting light signals into the module as electrical signals, and wherein the first and second mass storage modules are connected to the module bypass circuit board by a fiber optic transmission medium such that signals are communicated between the first and second mass storage modules in the form of light; and a controller managing a communication path between the first and second mass storage modules through the module bypass circuit board.

47. A high speed mass storage system which is readily expandable to increase its storage capacity while the system is in operation, comprising: ,04928114 31 O first and second mass storage modules, each mass storage module including at Sleast one storage device and at least one bypass circuit board associated with each d storage device; a module bypass circuit board including an optical input/output connector for outputting electrical signals from the module bypass circuit board as light signals and for 0O ID inputting light signals into the module bypass circuit board as electrical signals, and Swherein the first and second mass storage modules are connected to the module Sbypass circuit board by a fiber optic transmission medium such that signals are Scommunicated between the first and second mass storage modules in the form of light; and a controller managing a communication path between the first and second mass storage modules through the module bypass circuit board.

48. A high speed mass storage system which is readily expandable to increase its storage capacity while the system is in operation, comprising: a plurality of mass storage modules; a module bypass circuit board including an optical input/output connector for outputting electrical signals from the module bypass circuit board as light signals and for inputting light signals into the module bypass circuit board as electrical signals, and wherein the plurality of mass storage modules are connected to the module bypass circuit board by a fiber optic transmission medium such that signals are communicated between the plurality of mass storage modules in the form of light.

49. The high speed mass storage system of Claim 48, wherein said optical input/output connector comprises an optoelectronic transceiver. A high speed mass storage system which is readily expandable to increase its storage capacity while the system is in operation, comprising: C04928114 32 O first and second mass storage modules, each mass storage module including at N least one storage device and at least one bypass circuit board associated with each storage device; means for inputting and outputting light signals, said means for inputting and 5 outputting light signals outputting electrical signals as light signals and for inputting light oO ID signals as electrical signals, and wherein the first and second mass storage modules Sare connected to the means for inputting and outputting light signals by a fiber optic Stransmission medium such that signals are communicated between the first and second Smass storage modules in the form of light; and means for managing a communication path between the first and second mass storage modules through the means for inputting and outputting light signals.

51. The high speed mass storage system of Claim 50, wherein said means for inputting and outputting light signals comprises an optoelectronic transceiver.

52. A high speed mass storage system which is readily expandable to increase its storage capacity while the system is in operation, comprising: a plurality of mass storage modules; means for inputting and outputting light signals, said means for inputting and outputting light signals outputting electrical signals as light signals and inputting light signals as electrical signals, wherein the plurality of mass storage modules are connected to the means for inputting and outputting light signals by a fiber optic transmission medium such that signals are communicated between the plurality of mass storage modules in the form of light.

53. The high speed mass storage system of Claim 52, wherein said means for inputting and outputting light signals comprises an optoelectronic transceiver.