CA1180459A - Distributed data processing in ring-structured networks architected for full duplex peer-to-peer operation of processing stations and uninterruptible transfer of long data records between stations - Google Patents

Distributed data processing in ring-structured networks architected for full duplex peer-to-peer operation of processing stations and uninterruptible transfer of long data records between stations

Info

Publication number
CA1180459A
CA1180459A CA000417820A CA417820A CA1180459A CA 1180459 A CA1180459 A CA 1180459A CA 000417820 A CA000417820 A CA 000417820A CA 417820 A CA417820 A CA 417820A CA 1180459 A CA1180459 A CA 1180459A
Authority
CA
Canada
Prior art keywords
ring
frames
station
frame
information
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Expired
Application number
CA000417820A
Other languages
French (fr)
Inventor
Charles S. Lanier
Hiram M. Maxwell
Roger E. Mckay
Leonard Weiss
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
International Business Machines Corp
Original Assignee
Charles S. Lanier
Hiram M. Maxwell
Roger E. Mckay
Leonard Weiss
International Business Machines Corporation
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Charles S. Lanier, Hiram M. Maxwell, Roger E. Mckay, Leonard Weiss, International Business Machines Corporation filed Critical Charles S. Lanier
Application granted granted Critical
Publication of CA1180459A publication Critical patent/CA1180459A/en
Expired legal-status Critical Current

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F15/00Digital computers in general; Data processing equipment in general
    • G06F15/16Combinations of two or more digital computers each having at least an arithmetic unit, a program unit and a register, e.g. for a simultaneous processing of several programs
    • G06F15/163Interprocessor communication
    • G06F15/173Interprocessor communication using an interconnection network, e.g. matrix, shuffle, pyramid, star, snowflake
    • G06F15/17337Direct connection machines, e.g. completely connected computers, point to point communication networks
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L12/00Data switching networks
    • H04L12/28Data switching networks characterised by path configuration, e.g. LAN [Local Area Networks] or WAN [Wide Area Networks]
    • H04L12/42Loop networks
    • H04L12/427Loop networks with decentralised control
    • H04L12/433Loop networks with decentralised control with asynchronous transmission, e.g. token ring, register insertion

Abstract

DISTRIBUTED DATA PROCESSING IN
RING-STRUCTURED NETWORKS ARCHITECTED FOR FULL
DUPLEX PEER-TO-PEER OPERATION OF PROCESSING
STATIONS AND UNINTERRUPTIBLE TRANSFER OF LONG
DATA RECORDS BETWEEN STATIONS

Abstract Of The Disclosure In a ring-structured data communication network, in which plural data processing systems exchange data and control information on a full duplex peer to peer basis, systems are presently architected to assign at least three I/O subchannels (i.e. at least three device addresses) to respective ring interface adapters. At least two of these subchannels are dedicated for providing separate input paths from the ring to at least two associated program-assignable areas in their systems main store, and a third of these subchannels is dedicated as an output path from the system's store to the ring. Collectively, these subchannels can sustain two input transfer processes and one output transfer process concurrently. One of these input processes is associatable with a locked mode of adapter Operations which provides a non-blockable path for data transfer from a selected (remote) station on the ring to the respective system's main store.
In this mode the other input subchannel and the output subchannel permit the system to maintain full duplex communication with other ring stations in respect to network transactions/processes which may require priority attention. Information is sent on the ring in discrete information frames of variable bit length, each frame acknowledged by a response frame. In general, response frames have much shorter bit lengths than the information frames, enabling the systems to access the ring more efficiently than they would be able to if response and information frames had equal lengths.

Description

`9-81-020 DISTRIBUTED DATA PROCESSING IN
RING-STRUCTURED NETWORKS ARCHITECTED FOR FULL
DUPLEX PEER-TO-PEER OPERATION OF PROCESSING
STATIONS AND UNINTERRUPTIBLE TRANSFER OF LONG

_ckground OE The Invention This invention relates to methods of conducting data communications in ring-structured distribu-ted processing networks. In particular the invention concerns a method of operating processing station systems in such networks in a locked receiving mode~whereby a selected ring station may be given virtually continuous and exclusive access to the (locked) receiving station's system for transferring a variable amount of data into a program-assignable area of that system's main store.

A problem in res~ect to such operations is that a locked receiving station could be isolated from ring stations other than the selected station for intervals of time which are unacceptably lony, "unacceptably" in the sense that the benefits derived from locked mode ope~ation ~ay be ef~ecti~ely cancelled by the network disruption caused by the isolation of the locked station from other stations requiring immediate controlling attention.

The object of the present invention is to provide a method of operating such ring station systems in the locked mode which allows the locked system to remain eEfectively accessible to all ring stations.

Summary Of The Invention In accordance with the present invention, data processing systems operating in such riny networks are architec-ted to assign at least three I/O subchannels (i.e. at least three different device addresses) to respective ring interfacing equipment termed ("adapters"). At least two of these subchannels are dedicated as input paths from the ring to associated program-assignable (i.e. variable~ areas in the respective system's main store and a third of these subchannels is dedicated as an output path from that store to the ring.

These subchannels may be "armed" (i.e. readied by system programs to conduct respective input and output information transfer operations) independent of each other (i.e.
asynchronously) so that while a station is operating in a locked receiving mode through one of these input subchannels the other input subchannel and the output subchannel effectively provide the respective system with full duplex access to the ring for exchanging urgent priority communications with other ring stations.

In the present system data and control or attention request messages are respectively sent out on the ring in discrete data frames and request frames of limited bit length which are individually acknowledged by response frames returned to the data/request originator from the data/request destination station~ Data and request frames contain at least 7 bytes and not more than 1,007 bytes (1 byte = 8 bits), whereas each response ~r r ~

'9-81-020 -3-frame invariably contains only 7 bytes. Due to the generally shorter lengths of the response frames the ring is effec-tively more accessible for transmission access than it would be if all frames had equal durations.

In accordance with the in~JentiOn a station ring adapter is placed in the locked reception mode by one of two methods. In one method the adapter is conditioned to the locked mode by programmed commands performed by the respective system processor and channel. In this mode -the adapter stores the identity of a selected origin station to which it is being locked (a function accompanying the command signals) before any data has been sent by the origin station. In a second method the station adapter reacts to control information contained in a first frame of a plural-frame data message and enters the locked mode.
The control information indicates additional frames will follow. The adapter stores the identity of the sending station (also contained in the first frame) and begins locked mode operation while the first frame is being received.

With either method, the subsequent station/net-work operation is the same. The adapter acceptsdata frames directed to its station address (destina-tion) from the locked origin station and refuses similarly directed data frames from other ring stations. Accepted data frames are passed through one of the dedicated input subchannels to a main storage space prepared for that subchannel (by system programs) and an ac~nowledgment response frame is returned to the locked ori~in station permitting it -to transmit another data frame.
Refused data frames are discarded by the adap-ter and a "frame refused" response frame is sent -to the origin station indicating in e.E~ect tha-t the destination station was not prepared .Eor receptlon of such data f~amesO

Incoming request Erc~les are accepted by the locked adapter, passed through the other dedicated input subchannel to a main storage space prepared for that subchannel, and an acknowledging response frame is returned to the request/control source s-tatlon. The information in the stored request frame can be processed by the station system, and that system can send associated data or control to any station through the dedicated ou-tput sub-channel and the ring (in a data o.r request frame) while receiving data in the locked mode.

Accordingly, in locked mode plural frames of data sent by one station are exclusively received and stored by the destination station (without the potential disorder which would result if data from several stations could be passed to one subchannells storage space) and if the .Locked data trans~er occupies many frames it cannot block reception and processing of important reques-t frames.

For a more complete understanding of the in-vention, as well as a comprehension of other advantages and features thereof, reference should be made to the following description taken in connection with the accompanying drawin~s, and to the appended claims which indicate the scope of the invention.

~9-81-020 -5-Brief Descrip-tion of the Drawings Fig. 1 schematically illus-trates ~ ring network configurable for sustaining peer-to-peer data communications in a Eull duplex uninterruptable (locked) mode, in accordance with the present inventioll.

Fig. 2 illustrates various frame formats employed in the ring structure of Fig. 1.

Fig. 3 illustrates logical details of ring adapting equipment, embodying the operational method of this invention, for interfacing between a station processor and the ring.

Figs. 4-7 illustrate station/network processes for conducting communications between stations on a ring, in locked and unlocked modes, in accordance with the present invention.

Fig. 8 illustrates the relative timing of signal frames entering and leaving a typical ring station A; and Fig. 9 illustrates the sta-te oE occupancy of station A's FEQ buffer at various siynal flow stages shown in Fig. 8~

-~9-81-020 -6-Detailed Description Fig. 1 illustra-tes a ring network structured for enabling multiple data processing stations to conduct f~lll duple~ communications as peers (i.e.
without a central node or master station). Each station contains riny adapting equipment RCC
standing for (Ring Communications Controller) and a host data processing system. Each RCC in-terfaces between the ring medium R and the respective host 10 System, For the sake of simplicity, only four stations (A, B, C and D) are illustrated. It will be understood, of course, that a much larger number of stations could be accommodated.

The stations transmit information on the ring R in discrete frames having predetermined header formats and variable but limited bit lengths. A
single data record may occupy multiple frames.
The frames circulate unidirectionally on the ring -- clockwise in the illustration of Fig. 1 --with origin address, destination address and type specl~ler ln~ormation enabling destination s-tations to receive and selectively process the information.

The three presently pertinent types of frame formats, shown in Fig. ~, are "data", "request"
and "response". Data type frames con-tain data information, request type frames contain attention request or other control information, and response type frames contain information signifying accep-t-ance, refusal or incomplete receipt (due to error) of speci.fic data and request frames, Each data and re~ues-t frame must be discretely acknowledcJed, refused or signiied as incompletely ,received by a response Erame Erom the des-tination sta-tion, before the or:igin station can send another data or rec~uest frame (or re-transmi-t an incorrec-tly received frame) to that destinati.on.

As shown in Fig. 2, data and request type Erames have variable byte (bit) lengths -- in the present embodim~nt not more than 1,007 bytes (8,056 bits~ and not less than 7 bytes -~ wnereas response frames have fixed lengths of exactly 7 bytes. Each frame contains a four byte header part and a three byte end part. The header consists of a "sta.rt flag"
byte, "origin" and "destination" address bytes, and a "specifier" byte. The end part consists of two cyclic redundancy check bytes (CRC) and an "end flag" byte.

The CRC bytes are used for error detection/correction purposes not relevant to the present invention. The start and end flag bytes respectively indicate the beginning and end of the frame. The address bytes indicate addresses of the origin station which sent the frame and the destination station (or stations) to which the frame is being sent. The specifier byte specifi.es the frame type (data, request, response~ and certain other control information discussed later.

~5 The origin and destination of each response frame invariably correspond to the destination and origin of one and only one p,reviously transmitted data or request ~rame (sincel as noted previously, BC9~81-020 ,C9-81-02Q -~-each data and request frame must be discretely "acknowledged" by a response frame before another frame can be sent between the same origln and destination s-tation pair).

Fi~. 1 S}IOWS a station organization at A repre-sentative of the organi7ations oE all ring stations (A--D). Each station contains ring adapter equipmerlt 1 6 and a host processing system 7. The ring adapter con-tains receiving circuits 1, transmitting circuits 2, an FEQ (front end queue) buffer 3 for storing ring traffic in transit through the station to downstream destination stations, "in" and "out"
buffers 4 and 5 for respectively storing incoming ring traffic (or inspection and internal routing) and outgoing local origin information and response frames, and inspection control logic 6 for examining local destination informa-tion stored in the "in"
buffer and determining its handling. Each host system 7 contains a (central) processor 8, a (main) store 9 and an I/O channel 10 capable of sustaining multiple subchannel transfer processes concurrently (relative to the ring and other "peripherals").

Logic 6 examines inco~ g frarne headers, routes frames having local des-tinations to the in buffer, checks (and, if posslble, corrects) frames having local destinations, and routes frames havin~ remote destinations to the transmitter 2 via the FEQ. In respect to data and request frames having local destinations, a response frame is prepared and directly passed to the ring (via Out Buffer 4 and transmission equipment 2). The response frame distinguishes correct or incorrect reception of the respective data or request frame, and accep-tance or refusal of the frame (refusal if an input subchannel is not presently available for transferring the frame information into host main storage).

~C9-81-020 -9-In general, information contained in incoming data or request frames having a local destination is staged in the station in buE~er ~ and then conditionally transferred through an input sub-channel to host storage 9. Such stored inEormationis thereafter cluly processed by the station processor ~. Since a single clata message or record may occupy multiple data frames (each separately sent and acknowledged), processing of -the data contained in such messages is generally not initiated until after -the last frame has been received and stored.

The in buffer, ou-t buffer and "through-traffic" FEQ buffer 3 are each designed -to hold at least a maximal length data frame including he~der (i.e. a-t least 1,007 bytes). The FEQ is organized for first-in first-out (FIFO) operation and each buffer may be implemented by means of a RAM trandom access memory) storage array. As is usual for FIFO systems the FEQ requires in counting and out counting facilities for tracking locations at which information is to be respectively entered next and removed next.

Stat:ion bit transmissions are timed with reference to internal clock sources and station bit receptions are timed with reference to clock functions derived from incoming traffic (i.e. with reference to transmission clocks of preceding upstream stations). The station transmission clocks nominally have identical frequencies but need not be in phase synchronism. Accordingly, the station clocks may drift relative to each ot~er (over many frame transmission periods), requiring certain adaptation procedures. These procedures are not presently relevant, but they are described briefly below to illustrate an operating ring environment in which the invention may be advantageously used.

~C9-81-02Q -10-When a sta-tion's FEQ is not emp-ty the contents of its FEQ are passed directly to its output 2 in a FIFO sequence. When the FEQ is empty the s-tatior either transmits frames containing local origin data, contro:l or response inEormation, if such in-formation is available for transmission, or it transmits idle characters which are readily dis-tinguishable ~rom frame characters (not being bounded by start and end flag bytes). In order to allow for timing diferences between incoming and outyoing traffie the occupancy tag of a station~s FEQ is set from empty to not empty only when the FEQ contains at least four bytes and it is reset from not empty to empty only when the FEQ contains no by-tes (in count = out count). Consequently, while a frame is being passed through the FEQ to a station's transmitter there cannot be a "short fall" eondition (buffer empty before transmittal of the end flag of a frame partly transmitted) since at least four bytes of the frame must have been in the FEQ before the transmission began and the timing of a frame is short by comparison to the interval over which the transmissio~ clock could drift by the equivalent of four byte periods.

When a station is receiving idle characters from the ring it uses such characters -to maintain reception synchronism but does not enter them in its FEQ and does not advance its entry position address count for the FEQ. Consequently, if the FEQ becomes or is empty and idle characters are being received the station is permitted to send its local origin information while it is receiving the idle characters. This element of idle time usage, ~C9-81-020 coupled with the present employment of rela-tively short response frames, gives the stations on the ring earlier -transmission access to the ring than they would otherwise have and is considered a feature of the present invention.

The fourth/speci~ier byte of the response type frame (see Yig. 2) may indicate severa:L
different types of responses: a positive acknow--ledgment response indicating successful reception, a negative acknowledgment response indicating error in the received frame or a refusal response indicating rejection of the associated frame because the receiving system presently does not have an input subchannel available for transferring the 1~ frame information to its host storage (this might indicate a procedural error in preceding control frame communications in certain circumstances).

Another aspect of the present invention concerns the logical organizations of processor 7 and adapter controls 6 for handling data received in "locked"
mode. In accordance with the invention a station receiving a data message more than one frame long may be conditioned to operate ln a !~ locked" mode in which an input subchannel of the sta-tion's host ~5 system is dedicated exclusively for transferring that data message (into a program-prepared space in the receiving station's host storage), and (local-destination) data frames received from other tihan the selected station are rejected with a frame refused response. This mode of operation can be instigated either by the station's processor and I/O channel~ under program control, or while receiving the first frame of a plural-frame data -12~

message in response to control information ln the speci.fier byte of the fixst frame indicating tha-t other data frames will follow from the same station (usually as part of one conti.nuous message unit)7 Furthermore, each host processi.ng system in accordance with this inventiorl is architected to dedicate at least two of its input subchannels and at least one output subchannel for separately conducting communications between its main store and the ring. In other words, each station system is architected to assign at least three device addresses to its ring communication adapter, at least two of these addresses exclusively for conducting input transfers from the ring to its host store and one address for conducting only output transfers to the ring. This architectured ded.ication feature guarantees availability of full duplex linking facilities between the station and the ring for co.nducting control communication~ ~receiving request frames from the ring and sending request or data frames out on the ring) while one of the dedicated inpu-t subchannels is operating in the foregoing loc]ced mode relative to receiving a plural frame data message. It thereby guarantees that time-critical action required between two or more stations in the network cannot be blocked indefinitely by a locked mode data receiving operation in one of these stationsO

Fig. 3 illustrates the logical organization of a subject station system for operating in this locked data receiving mode. The logical configuration of the station when not operating in the locked mode will be explained with reference to this Figure and Figs. 4 and 5. ~tation system 3~ procedures for entering (and leaving) the locked mode will be explained with reference to this ~igure and Figs. 4, and ~.

~C9-81 020 ~" ;"

Referring to Fig. 3, incoming rin~ frames are passed through receiver 100 and path selector 101.
Receiver 100 demodulates the received signal, extracts bit timing reference clocking from the demodulated si~nals, and passes the demodula-ted bîts to selector 101. Selector 101 distingulshes idle chAracters from frame characters (recall that frames follow a start flag byte contiguously, as shown in Fig. 2~ whereas idle characters will contiguously trail end flag bytes), discards bytes which represent idle characters, dis-tinguishes frames having a local destination from other frames, passes other frames to -the FEQ buffer 102, and enters local-destination frames into the In Buffer 103~ Frames having a "broadcast address"
destination are steered into both the F~Q and the In ~uffer (such frames, which are usually request type frames intended for a group of stations or all stations, circulate through all stations and are removed from the ring at the origin station when they return to that station via the ring).
Selector 101 includes a not-shown delay for holding the first three bytes of the incoming frame header until the routing decision is made.

The FEQ contains a not-shown random access storage array (R~l) and a not-shown pair of "in" and "out" counters or respectively indicating next addresses in the ~P~I at which incoming bytes are to be entered and removed. When the FEQ becomes empty (in count = out count) an empty condition is mani-fested on line 104. When this condition is mani-fested and the FEQ acquires four ~ytes (in count -out count = ~) the indication on line 104 is changed to manifes-t a not empty condition. This delay in manifesting the not empty condition is used to ensure that transmission of information from the FEQ

to the station's output ring port will not begin prematurely a~d thereby potentially crea-te an overrun condi.tion because of tlming difEerences between the s-tation's separate -transmission and reception clocks (the :Eormer derived from the received si~nal.s and the latter generat.ed separately from an internal crystal oscilla-tor)~

Transmission source selector 105 reacts -to manifestations on line 104 and other control indications noted below to select information for application to the station's transmitter 106 from either the FEQ or the station's Out Buffer 1070 When the FEQ is empty and the Out Buffer con-tains either a complete response frame in its response area 107a or a complete locally origina~ed data or request frame in its local stagin~ area 107b, a frame is transferred from the Out Buffer to transmitter 106 (if areas 107a and 107b are simultaneously full the contents of 107a are transferred first). If the FEQ is not empty, and a frame is not currently being transferred from the Out Buffer to the transmitter, the FEQ is unloaded to the transmitter on a FIFO basis. If the FEQ is empty and the Out suffer does not contain a complete frame the selector 105 causes idle characters to be sent by the transmitter. When such idle characters are received at the next station on the ring they .~re discardecl by its input selector 101.

Receiver input selector 101 applies (local-destination) data and request frames to In ~uffer 103 via bus 108. Incoming (local-destination) response frames are used to operate not-shown indicators for directly signalling to the host system channel the status oE completion or non-completion of any previously sent local-origin data or request ~ommunications.

BC9~81-020 Frames are transferred conditionally from the In Buffer to output bus 109, and from tha-t bus to the hos-t system's t/O
channel 110, depending on the availability of a su:i-tably prepared input subchannel as described next.

Channel 110 contains plura:l subchannels lll. ~s explained previously, at least two of these subchannels (only ~wo in the presently disclose~ embodiment) are dedicated e~clusively as input subchannels for conducting input transfers from the ring R to host storage (via the In ~uffer), and another one of these subchannels is dedicated as an output subchannel only for conducting output -trans:Eers from host storage to the ring (via the Out Buffer). The dedicated subchannels are designated herein as subchannels 0, 1 and 2 ~abbreviated SC0, SC1, and SC2~. SC0 and SC1 are dedicated as input subchannels and SC2 is the dedicated output subchannel. Furthermore, in the presently disclosed embodiment of the invention, SC0 is dedicated exclusively for inputting information contained in request frames and SC1 is dedicated e~clusively for inputting information contained in data frames. Other subchannels 111 are used for sustaining communications between the host system and its "peripherals" via a bus indicated generally at 112.

The dedicated input subchannels are prepared for conducting respective input information transfers as follows. With reference to the host main storage map indicated at 113, when SCl is available and required (for an input data transfer) application programs at 113a allocate a space 113b for data storage and prepare a command at 113c for "arming"
SC0. When this command is performed (by channel 110) SC1 is "armed" (readied for immediately -16~

transferring data from In BufEer 103 to space ]13b) Similarly, when SC0 is available and needed, host applicatlon programs prepare a Read Request/Control Information command at 113d which when performed "arms" SC0 for -transEerring inEormation contained in a request frame from the In Buffer to a prepared area 113e in host storage.

For output transfers, host programs load the outgoing information into a selected area 113f and schedule a Write (output) command shown at 113g for execution when SC2 is accessibleO The write command arms SC2 for conducting an output transfer of one or more frame-loads of information between the area 113f and the Out Buffer 5One frame-load at a time, and each frame-load after -the first conditional on prior receipt of a positive acknowledgment response from the destination station).

Lines ll~ 116 represent control signalling paths for respectively arming SC0, SC1 and SC2. SC0 and SC2 are armed simply by applying signals to these subchannels for transferring them from "unprapared" to "prepared" states.
In their prepared states SC0 and SC2 are readied respectively for immediately inputting request information and outputting outgoing information. SCl is also armed in this manner, but in addition, SC1 may be armed for operation in either an unlocked or locked mode and receives a conditioning signal es-tablishing one of these modes.
Furthermore, when operating in the locked mode SC1 receives a station address representing its exclusive origin for input dataO In the locked operation input data from any other source is refused.

, ~, Internal path selector 118 dete.rmines whether infvrmatiorl containecl i.n local destination incoming frames ~in the In Buffer) should be refused or passec1 ~to host storage via SC0 oY SCl). The selector also selects a suitable response ("positive error-free acknowledgment", "negative error-associated acknowledgment" or "frame refused") and applies an appropriate signal to transmission path selector 105 causing the latter circuit to generate a corresponding response frame which it transfers to Out Buffer area 107a for transmittal via the ring to the incoming frame's origin station. Signals received via lines 119-122, and s-tored by the selector, determine its action.

The action of selector 118 is conditioned on the incoming frame type (data or request), the preparational state (prepared or unprepared) of the associated subchannel (SC0 for request~ SCl for data~, and in respect to data frames, the locked or unlocked status of SCl. If a request -type frame is received and SC0 is prepared the frame is passed to SC~. If SC0 is not prepared, the frame is rejected (discarded) and a frame refused response is returned to the origin station. If SC0 is prepared and receives the frame without error, the frame is stored in program-prepared area 113e (of host main storage) and a positive acknowledgment response is returned to the frame's origin station (via selector 105, the Out Buffer, transmitter 106 and the ring).
If the frame contains an uncorrectable error an error indication is stored in SC0 and a negative acknowledgment response is returned to the origin station.

~C9-8~-020 ~9-81-020 -18-If the frame is a data type frame and SCl is prepared and unlocked the frame is simply passed through SCl to program-prepared storage space 113b and a positive or negative aclcnowledgment response is returned to the sender station depending respec~
tively on whether -this operation is completed without or with error. In the same circumstance if SCl is not prepared the frame is rejected and a rame refused response i5 returned to the sender.
If SCl is both prepared and locked the action of selector 118 depends on the frame's origin, as represented by the third byte (origin byte) in its header. Selector 118 receives this address via linè 120 and compares it to the origin address which it received from line 121 during the preparation of SCl (and then stored). If the compared addresses match, the frame is accepted and the data is passed to SCl. A positive or negative acknowledgmenr response is returned to the sender depending on the state of completion of the transfer ~without or with error), and if the transfer is successful the data is stored at 113b. If the compared addresses do not match the frame is rejected and the frame refused response is returned.

In respect to the foregoing transfer of data when SCl is prepared and unlocked, selector 118 - conditionally may establish a locked mode of opera-tion during this type of transfer. Selector 118 receives specifier byte information, via line 119, indicating whether more data frames will follow from this frame's origin (usually, as part of one continuous data message or set and in association with one subchannel preparational command). If ~C9-81-0~0 -19-more data will not follow (i.e. this is the only data Erame being communicated) the frame is simpl~
passed to SCl as above. However, if the speclfier byte indicates more frames to follow, selector 118 stores the frame's origin address (received via line 120) and condltions itself ancl SCl for locked mode operation as described above~.

The foregoing operations are summarized in the followin~ table.

~ 459 ~ ~ l~
rJ C) ~ a h O ~ ~ rl aJ ~. ~ ) h u r~ Q O O Orl (I) ~1 0 tl)P I h .1-tn '7~ 0 rl ~ ~a t~ ~ o r~
~ dP~ (u a a) a~ ~ u) ! I rl(1.) U~ r~
p~ ~ o u~ ~ h U~ D h O o r: a h u t) r.~
~ O ~ ~ ra -~
o ~ 1 a) tn o r~
O O ~ Ul r~ ~- rl O ~S ~1) (1) ~a r I a) 1~ ~r-l h u u~ C:
O V Ir~ (I) r~ h a) ~ h O F ~ n ~ u) o h O
O h O O ~ ~rJ p ~a ra ~ n~ ~ o r~
a) ,~ Q,~ r~
o x ~ ~ a~ Ul ~: h ~; ~ OaJ O ~1) O ~ rl h t:~' X 0~ ---~; h r~ O ~ X
Z ~ ~u O ~ h t~ : h ~ ) O
O E-l ~ rJ O ~ ~ ~d O h ra ~1 F~ Z ~ d ~ r~
~ O ~ JJ . ~ ~ d a~
Z H ~a.)U7 U~ ~ O
H ~ OU~ 0 11) t~ ~ r~
E~ ~ r~ J h U~
¢ ~ ~ - rl O ~)a) ~ O U~
r~ O ~ -r~ ) O 11) Q rl rl ~ ~a) E~ ~) ~ ~' ` ~ ~ O
r~ ~ O r~ ~a r~ ~ ~a a~ h ~¢ c ) ~ ~ a) ~
~ ) U~~q ~rJ ~ O :u~ h o ~~a) r~(r~l 1~d : O h ~ r~ r~ 1~1 h Z; '~.~, h 0 4 ~:
H tl) ~) ~U h ~ O'd o ~ ~a ~d ~
U~ O h a) r~ O a) h rd ,~ U~ h U~ h O
X i~ a~
R~ h ~~r; U~a) (I) h U~
~n rd v al a) ~ hS~ ~ h ¦~ ~ ~,) ~J ~rlR~ O O ~ U~ ~ ~
~\rl ~1 ~ rd R. ~ RJ ~ ~ h R~ I
~a o ~ h ~ ~a ~ rd O ~u~ a~
r~ l r~ ~~ v ~ h ~ ~ u o,~~n a) a) a) o a a ~ r ~ ~
r~ r~ ~ nt 1 J (~) ~J
~> h>~,) rJ~ h h O h rd ~ ~a H r.~ R O ~ U~ r~ r- ~ r~ ~ rL,:~ -- a r~

r~ -~
~, a) a~
H r~ ~ ~ r ~ ~; ~1 ~ O ~ O
a :3 ~ ~ ~ V .1_~
r~ E~ a a ~ a --a) r; $-1 r~ ra a) h ~ h r~ r~ td IV X X rd ~ ~ a) ~ al ~
E~ c,q h 1:: h h r,q r~ r~
~; ~ ,_ ~tl ~
O ~ h t~ rd ~:: ~ o rd X td a) X X
r~ o R~ h m _, ~ R~
::~ S~ ,~
,.q X ~ ~
o "~ ~a 'd r~
O Y X X X ~d ~ ~a ~ r~l ~1 ~ ~.) O O V O u~ q co E~ a o o o o X X
~ O ~ u o u o a~
E~ ~ ~ O ~ O ~ C 3 ~n ~ ~ ~ ~ ~ ~

a) ~ o o ~: ~ a tn u~ ~ o o o Q~ ~ O
u~ ~ ~

u ~ u 0 ~ ~
` o ~ s~
a) u,r~
~: a) ~ a Z u~ ~` Q, u~
~1 5 ~ O O~; rl ~ ~
f~ ~ ~, ~ Q.
E~ .Y h u t) o a) ~ ~ ~ ~I ~ O O
O
H U~ a ~
E-1 dJ a.) ~ ~
C ) Q~ ~J u~
.,~ ~ O
~ a~ h ~ 11~
h O a)o u~ Q, O ` ~ ~o t~ Id S~
t~ O h o u~ h -i ., ~ u~ a~ ~ aJ
o h ~ o ~ ~ o U~ ~ ~~ td ~
O a) ~ h o ~ ~~ h U~
~1 h h,1 1~ h rd . o Q, ~:1 0 a) o u~ o ~ ~ ~ o u~ ~ h U~
,a ~ h~1 h ~
l q ~ I
~1 a1a.) a~ a O ~ ~ O
~ 1 h ~--I ;-~ . . r1 .~ `1 n ~ ~ ~4 ~U ~4 h ~4 Q

, o U~

a) a~
~Y; ~ ~ aJ ~ a~
X
O ~ O
S~
~4 E~ ~ U m ~; m ~;

~ U X X X X ~x E~ ~n cn ~ a) æ ~ h ~a SJ
æ
O
X ~ ~
~,3 u~ ~ h ~ h m ~ Q. ~ ~
u~ ~ 3 P- ~
o o ~; ~) I
O ~ ~ U~ ~
H 1:~:1 a) C3 U) CO
E~ a ,~ ~ ~ x x f~C O u u Q.
E~ ~5 O O
U~

In order to generate the response Erame selector 105 takes the origin address byte of the incoming rame's hea~er and positions it as the destination hyt:e of the response frame.
It also takes stored constants representlng -the local station addxess, start flag and end flag functions, and inserts these into appropriate response frame positionsO It generates a specifier hyte as a func-tion of the response type (positive ack, negative ack or frame refused~ and appropriately positions that byte in the response frame.
Finally, it generates CRC bytes as a function of other bytes in the frame. The Out Buffer, as previously mentioned is a RAM array having areas 107a and 107b respectively dedicated for response storage and outgoing local-origln message storage. The response frame bytes are entered into appropriate positions in area 107a as they are generated, and when the complete response frame has been stored a signal is given to transmitter 106 indicating its availability for transmissionn The foregoing table indicates a broadcast frame type not previously mentioned, and not particularly relevant to this invention. The broadcast frame is a request type frame intended for reception at more than one station, either a particular group of stations or all stations on the ring.
When such a frame is received and SC0 is prepared the frame is passed to SC0 but no response is generated. The frame is also steered into the FEQ by selector 101 while it is being placed in the In Buffer. Accordingly, the frame will circulate through the FEQ to the ring output and pass downstream to the other stations on the ring. When the BC~-81-020 -23-frame re-turns to its origin the selector 101 at that station will recognize its origin and remove it from the ring (suppress it). IE the frame is received at a station not having its SC0 prepared it is merely "repeated" via the station's FE~. Since such frames are not ac~nowledgecl it is incumbent on any statio-l having a messacJe to send in this broaclcast mode to send the messacJe several times in order to ensure its reception at all intended stations.

The above table also indicates a "bypassed"
mode of station operation not presently relevant but deserving brief mention. In bypassed mode the station's receiver output is coupled directly to the station's transmitter input bypassing the s-tation's selectors, F'EQ and internal paths. This mode permits the station's internal logic to be isolated from the ring for error recovery purposes.
As noted in the table, in this mode all frames directly bypass the station, whereby any frame intended for local reception will return to the origin station with the same form as when it was transmitted and without a response. In this case the selector 1~1 a-t tne orlgln station recognlzes the occurrence, removes the frame from the ring (suppresses it3, and may also pass a signal to its host system for alerting host programs to the occurrence as an indication of an altered ring configuration.

The flow diagram of Fig. 4 provides an overview of how stations prepare for receiving locally-clirected request messages and how they handle such messages -thlough their SC0 subchannels. Block 140 indicates operations of hos-t system programs in respect to preparation of SC0. The hos-t's application and scheduler software prepare (reserve~
suitable storage space (i.e. 113e, Fig. 33 and schedule the execution of an I/O initiating instruction. Block 141 indicates that this instruction activates the channel to set up a read ~input) transfer operation, in "ring-dedicated"
subchannel SC0, which places SC0 in a prepared ("armed") condition. Blocks 142 and 143 indicate that with SC0 armed a request type frame sent by any ring station Istation C in the example) and having local destination, will be accepted on receipt and "immediately" passed to SC0 for input handling. Decision 144 and blocks 145 and 146 indicate that if the resulting input transfer is accompanied by an error a negative acknowledgment response will be returned to the sender and the channel operation will conclude with a report of error status to the host software (i.e. via an I/O
interruption). Decision 144 and blocks 147 and 148 indicate that if the input transfer is error-free a positive acknowledgment response is returned, the frame information (attention request or control message) is stored in host storage and the operation is concluded with a report of successful status to host software.

,.,.,~, ,,~ . i.

With either successful or unsuccessful conclusi.on SC0 is placed in the unprepared (di.sarmed) condition. If a local-destina-tlon request frame arrives while SC0 is disarmed the frame i 5 rejected with a frame refusecl respo}lse. AccordingLy, if the hos-t system software is supposed to minimize such refusals (in order to conserve ring banclwidth) the software should be designed to rearm SC0 quickly (as soon as possible a:Eter receiving the concluding status)O Of course, -this aspect of software operation is not relevant to the present in~ention.

Fig. 5 indicates how stations conduct their output transmissions, in an example describing transmission of a message from ring station A to ring station D. Blocks 150 and 151 indicate that host software prepares outgoing message information (attention request, control information or data) in a suitable space in host storage and associate the message with the device address of SC2 and with the "sub-address" of ring station D. Blocks 152 and 153 indicate that the station A channel forwards the information to the ring adapter which passes it to station D. If the information occupies more than one frame it is passed one frameload at a time. The channel arms SC2~ passes a frameload of information through SC2 to area 107b in the Out Buffer (fig. 3~ and waits for a response to return ~ia its ring adapter before either concluding the operation (block 154) or passing another frameloa~ of information to the adapter.

~C9-81-020 5~

In the presen-t embodiment only data messages may occupy more than one frame. When passing the flrst :Erame oE a plural-Erame data messa~e to -the r:iny adaptex the channel provides an indication to SC2 and the aclapter tha-t at leas-t one more additional frame will be sent in this operation.
SC2 then conditions the adapter to prepare a specifier byte for the first frame indicating that it is ~oth a data frame and not the last frame of the associated data message. As mentioned previously, if SCl at the receiviny station is armed and unlocked this information is used to set up a locked mode of reception while the first frame is beiny received. When the last frame is sent SC2 and the adap-ter are conditioned to provide a last frame indication in the outgoing specifier byte, enabling the receiving station to pass a corresponding indication to its host system and thereby permitting the receiving system to bring it.s corresponding read/input operation to an early conclusion.

Fig. 6 shows the process at a typical station (station A) for setting up a locked mode of "plural-frame" data reception operation relative to another station (station C) before that station sends any data frames. Another method for entering the locked mode, while receiving the first frame of a plural-frame data message in unlocked mode, will be discussed later with reference to Fig. 7. In this process station A receives a locking request frame directed from station C to station A (block 160), and application software in station Als host system prepares host storage spaces for subchannels SC0 and SC1 (block 161). The space prepared for SC0 is sufficient -to receive at least one request frame from any station on the ring, and the space prepared for SC1 may be sufficierlt to receive the data occupying a number of data frames specified in C's request.

BC9~81-020 Station A's software then issues an initiating instruction and a Read Lock command xelative to SCl, and a separate initiating instruction and a Read command relative to SC0 (blocks 162 and 163). The Read Lock command prepares SCl to receive (local destination) data contents of data frames from A' 5 ring adapter only lf sent from station C and to refuse data frames sent to A from any other station (block 164). The Read comrnand prepares SC0 to receive the request or control information in one request frame sent to A from any station.

Accordingly, any request frame arriving at A
during the reception and response process associated with C's data will be passed to A host storage and processed. If A must send information to another station while receiving C's data (e.g. if A must respond to a request) A's software will prepare the information in storage, arm SC2 and send the information in one or more data or request frames.
When the locked reception process is complete/
concluded SCl is disarmed. SC0 may be disarmed before or after this conclusion, depending upon the time of arrival of a request to A.

Fig. 7 indicates in a more comprehensive view the process of data reception at station A. This Fig.
shows the process for receiving data in a locked mode established either by host system software (in accordance with Fig. 6) or by the ring adap-ter it-self (in response to control information in a first frame of a plural frame data message).

5~

As each data frame having a local desti~ation is received in station A's In Buffer (block 170) --after various stages of ring adapter activity suggested at 171 by -the circled connection symbol "a" and at 172 by a simple broken line -- selector 118 in the ring adapter (Fig. 3) conditions the handlin~ of that frame on the armed or not armed state of data lnput subcllannel SCl (decision 173).
If SCl is not armed the frame is re~ected and a frame refused response frame is returned to the origin station (block 17~). The selector is then prepared (via sequence connections 175-171) for handling a next incoming data frame (block 170) or request frame (not shown), and SCl may be armed in a not-shown in-termediate process.

If SCl is armed at decision stage 173 the handling of the frame is conditioned on the locked or unlocked state of SCl (decision 176)~ If SCl is currently not locked the data is passed to SCl as shown at 177. If an error is detected (decision 178) a negative acknowledgment response frame is returned to the sender (block 179) a~d the selector becomes available (180/171) for handling another frame.

~ s indicated at 181, if the data is received by SCl without error a positive acknowledgment response frame is returned and the frame data is stored (in the space prepared for SCl in host storage). In this circumstance, selector 118 examines the specifier byte of the data frame and determines (from the state of a bit in that ~yte) if this is the only data frame currently being sent by its origin station or if other frames will follow (decision 182). If t~e frame will not be followed by other ~C9-81-020 -29-frames the operation is concluded (SCl is disarmed and, if necessary, re-armed), and no further ac~i.on is taken. However, if -the frame is to be followed by another data frame the selector and SCl lock to the orlgin station (block 183) for exclusively receiving data frames thereafter only from that station. In this operation the selector stores the origin station's identity and sets SC1 to locked sta-te.

Assuming that the "yes" decision had been made at 176 (SCl armed and locked), the selector would determine (decision 185) if the frame was sent by the station to which A is currently locked (by comparing the frame's origin address to the origin station identity which had been stored previously in the locking process). If the frame source does not match the origin identity the frame is refused (block 186). If the frame source does match the data is passed to SCl (block 188). If an error is detected in passage (decision 189) a negative acknowledgment response frame is returned (block 1~0). If the frame data is passed without error it is stored in host storage and a positive acknowledg-ment response i~ returned (as shown at 191~.

Fig. 8 illustrates the timing of frame signals entering and leaving a.typical station (station A) at its ring interfacel and Fig. 9 illustrates the condition of that station's FEQ buffer at various times specified in Fig. 8. Parts a-d in Fig. 8 indicate station input/output timing with the station equipment in various initial conditions and parts a-d in Fi~. 9 indicate FEQ occupancy conditions at times shown in respective parts of Fig. 8.

s~

At tO in Fig. 8a the station is about to receive a clata :Erame origina-ting at stati.on D and destined for station B, roll.owed in time by several iclle charac-ters, then a request frame in transit ~rom station C to station B, and then more idle characters. Since the station is not sending any locally originated information traffic at tO (Out Buffer not full) and since the FEQ becomes not empty shortly after tO
(i.eO as soon as ~our bytes of the incoming frame have been received) the station will begin to transmit the "D to B"
data frame after a four byte time delay relative to tO
(assuming that the Out Buffer remains not full while the four bytes are being recei~ed). The station will then continue, without interruption and without sending any idle characters, to send the "D to B" and "C to B" frames contiguously in time. At time tl, when the station is about to send the ].ast two bytes of the "D to Bl' frame, its FEQ
contains the last two bytes of the "D to Bll frame and the first two bytes of the "C to Bl' frame will be entered with the contiguous FIFO positions (see Fig~ 9a, and recall that the idle characters separating these frames at the station input are used only for reception synchronization and are not stored in the FEQj.

At time t2 in Fig. 8b it is assumed that station A has received a data frame directed to A from station B. To simplify the illustration, it is also assumed that the frame was preceded and followed at the receiver input by idle characters. As shown on the output line of this part of Fig. 8 the station hegins to send a corresponding response frame to station B after an indefinite de]ay (relative to t2) associated with the station's internal process for loading its In Buffer and passing its data contents 5~

to SCl (assumed -to be armed). The response frame could be a positive or negative acknowledgment (no error or error) or a frame refusal (if A is locked to a station other than B).
As shown in Fig. 9b, at t~ the E'EQ is in an empty condition (In Count = Out Count).

At time t3 in Fig. 8c i-t is assumed that station A has just finished transmitting a local origin message frame to station C (from its Out BufEer) and has received the first twelve bytes of a frame being sent from sta-tion D to station B. As shown in Fig. 9c, A's FEQ contains the first twelve bytes of "D to B" frame at time t3. As indicated on the output line part of Fig. 8c station A will begin to send the D to B frame immediately after sending the last byte of its A to C frame, and then begin to send a C to B frame (which follows the D to B frame on A's input~ as soon as it finishes sending the last byte of the D to B frame.

Finally, Figs. 8d and 9d indicate the handling of a local origin A to C transmission when A's Ou-t Buffer becomes full while the station is sending a D to B "through-traffic"
frame partially contained in its FEQ. At time t4 A's FEQ
contains four bytes of the D to B frame (see Fig. 9d) which A is then transmitting. It is also assumed that A's Out Buffer becomes full at or after t4 (with the information to be sent from A to C), and that the incoming ring traffic arriving at A after the D to B frame consists of a C to B
frame followed by a number of idle characters and then a D
to C frame.

~C9-81-02~ -32-Since the idle characters are being received when A
completes its transmission of the C to B frame, A
will i~nedia-tely begin to send its (local origin) A
to C frame after sendi.ng the last character of the C to B frame, and then con-tiguously send the D to C frame after that frame has been partially delayed in A's FEQ.

While there has been described what is at present considered to be a preferred embodiment of this invention~ it will be obvious to those skilled in the art that various changes and modifications may be made therein without departing from the invèntion. It is, therefore, intended to cover all such changes and modifications in the following claims as falling within the true spirit and scope of the invention.

Claims (14)

The embodiments of the invention in which an exclusive property or privilege is claimed are defined as follows:
1. In a ring-structured data communication network in which plural data processing systems are adapted to communicate with each other on a full duplex peer to peer basis, by circulating information around a ring to which said systems are serially linked, an improved method of conducting such communications comprising:

dedicating at least two input subchannels and one output subchannel in each system for passing information between the respective system and said ring;

conditioning each of said systems for operating selectively in locked and unlocked receiving modes when receiving information from another selected one of said systems through one of its dedicated input subchannels;

preventing information sent by a system other than said another selected system from entering the dedicated input subchannel of a system operating in said locked mode relative to said selected system;
and steering certain information received from systems other than said selected system through the other dedicated input subchannel of the system operating in said locked mode for enabling said locked system to receive and process urgent messages sent by systems other than said selected system, and to send information to any other systems via its dedicated output subchannel and said ring, while operating in said locked mode.
2. The information communication method according to Claim 1 including:

conditioning said systems to transmit information over said ring in discrete information frames, each requiring individual acknowledgment by its destination system, said frames having variable bit lengths not less than a predetermined minimum length and not longer than a predetermined maximum length;

conditioning said systems which are destinations of said frames to acknowledge reception of respective frames by transmitting associated response frames on said ring to the systems which respectively originated said information frames, said response frames having fixed lengths not exceeding said minimum length; and conditioning each said responding system to fill any time vacancies on said ring which result from the difference in length between its transmitted response frame and the associated received information frame by sending idle characters over said ring which enable a next downstream station on the ring to maintain its reception synchronism while also permitting that next station to originate information transmissions on the ring while it is receiving the idle characters.

3. In a ring-structured data communication network in which plural data processing systems are adapted to conmunicate with each other on a full duplex basis, by circulating frames around a ring to which said systems are serially linked, an improved method of conducting said communications comprising:

dedicating at least two input subchannels and one output subchannel in each system for passing information between a store in the respective system and said ring;

preparing one of the dedicated input subchannels in a first one of said systems for receiving a message exclusively from a second one of said systems via said ring;

sending said message from said second system in a plurality of discrete frames, each frame requiring a separate acknowledgment to be returned by said first system via the ring before the next frame can be sent by the second system;

preparing the remaining input subchannel of said first system for conditionally receiving information while said one input subchannel is being used exclusively for receiving said second system's frames;

examining incoming frames directed to said first system via said ring from stations other than said second system while said message is being sent by said second system; and
Claim 3 (Cont'd.) steering a selected one of said examined frames to said first system's store through said first system's remaining input subchannels.
A method of conductincJ ring communications in accordance with Claim 3 comprising:

completing said preparation of said one input subchannel in said first system for receiving exclusively from said second system before any frames are sent by said second system; and notifying said second system, via said ring, to begin sening its message.
5. A method of conducting ring communications in accordance with Claim 3 wherein said exclusive preparation of said one input subchannel in said first system comprises:

preparing said one input subchannel in said first system to receive information frames from an unspecified system; and when the first frame of said message arrives from said second system conditioning said one input subchannel to receive frames thereafter only from said second system.
6. The method of Claim 5 including:

incorporating control information in said first frame sent from said second system indicating that at least one more associated message frame will follow that first frame; and basing said exclusive conditioning of said one input subchannel on detection of said control information at said first system's interface to said ring.
7. The communication method of Claim 3 comprising:

preparing said one output subehannel in said first system for transmitting information to another system via said ring; and transmitting said information through said one output subchannel while said second system's message is still being transmitted.

8. In a ring-struetured data communication network, wherein multiple serially connected data processing stations inter-communieate on a full duplex basis as peers, a method of conducting communications between said stations comprising:

Claim 8 Cont'd.) receiving a continuous stream of uniformly timed incoming digital bit signals at each station, from a preceding station on the ring, and transmitting a continuous stream of uniformly timed outgoing digital bit signals from each station to a next station on the ring; said signal streams comprising discrete frame packets of information variably separated in time by bit groups representing idle characters; each said frame packet containing origin and destination information specifying ring locations of stations which respectively originated the packet and are intended to receive the packet; each said information frame packet also containing other control information;

examining said incoming signal stream at each station and removing therefrom all idle characters, as well as those frame packets which designate the respective station as either an origin or destination;

storing information contained in certain of said removed frames for additional processing, while discarding other said removed frames and said removed idle characters;

variably delaying portions of said received stream, excluding said removed idle characters and removed frames, within each station, by circulating said portions contiguously through a variable depth front-end insertion queue;
Claim 8 (Cont'd.) locally generating idle characters and information frames, at each station, for outgoing transmittal;

variously interleaving frames taken from the insertion queue with said locally generated idle characters and frames, at each station, and transmitting resulting composite signal functions on the ring as a said outgoing stream; and selectively processing certain frames removed and stored at said stations for variously establishing and dissolving exclusive locked relationships between pairs of stations; one station of each locked pair being conditioned for selectively storing information contained in incoming frames having the respective one station as a destination, but only if said frames originated at the other station of the respective locked pair, said one station being conditioned to discard certain other incoming frames having the same destination.
9. The ring communication method according to Claim 8 wherein said information in each frame comprises data, request or acknowledgment information, comprising:

including in said locally generated frames, at each station, acknowledgment frames corresponding to each remotely originated data and request frame having the respective station as a destination; each such acknowledgment frame containing an origin indication corresponding to the address of the respective removing station and a destination indication corresponding to the address of the station which originated the respective removed frame.
10. The ring communication method according to Claim 9 comprising:

arranging said acknowledgment frames to have substantially shorter bit lengths than corresponding acknowledged frames.
ll. The method according to claim 10 comprising:

arranging said acknowledgment frames to distinguish between corresponding acknowledged frames which have been removed and stored for processing at respective stations and corresponding acknowledged frames which have been removed but discarded at respective stations.
12. The method according to Claim 9 comprising:

allocating storage facilities at each station for separately storing data and request information contained in remotely originated data and request frames specifying the respective station as a destination;

selecting certain of the remotely originated data frames removed at each station for storage in the respective facility allocated for storing data;

selectively channeling the data contained in said selected data frames to the respective facility allocated for storing data; and channeling information contained in received request frames to the facility separately allocated for storage thereof.
13. The method of Claim 12 comprising:

reserving separate input subchannels for association with said separate storage facilities, and channeling information contained in said selected data and request frames to respective storage facilities via respective reserved subchannels, whereby the selection, storage and processing of such frames may be accomplished in a time-overlapped mode.
14. The method of Claim 13 comprising, at each station:

dedicating one of said reserved input subchannels for handling only input data transfers between said ring and said storage facilities and another of said reserved input subchannels for handling only request frame information input transfers; and conducting all input transfers of data only through said one input subchannel and all input transfers of request information only through said another input subchannel.
CA000417820A 1982-01-25 1982-12-15 Distributed data processing in ring-structured networks architected for full duplex peer-to-peer operation of processing stations and uninterruptible transfer of long data records between stations Expired CA1180459A (en)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
US342,439 1982-01-25
US06/342,439 US4489379A (en) 1982-01-25 1982-01-25 Distributed data processing in ring-structured networks architected for full duplex peer-to-peer operation of processing stations and uninterruptible transfer of long data records between stations

Publications (1)

Publication Number Publication Date
CA1180459A true CA1180459A (en) 1985-01-02

Family

ID=23341832

Family Applications (1)

Application Number Title Priority Date Filing Date
CA000417820A Expired CA1180459A (en) 1982-01-25 1982-12-15 Distributed data processing in ring-structured networks architected for full duplex peer-to-peer operation of processing stations and uninterruptible transfer of long data records between stations

Country Status (2)

Country Link
US (1) US4489379A (en)
CA (1) CA1180459A (en)

Families Citing this family (40)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPS5979655A (en) * 1982-10-27 1984-05-08 Toshiba Corp Data transmission system
US4667287A (en) * 1982-10-28 1987-05-19 Tandem Computers Incorporated Multiprocessor multisystem communications network
SE435987B (en) * 1983-06-09 1984-10-29 Ericsson Telefon Ab L M SET AND DEVICE FOR CONNECTING A CLOSE RING THROUGH A PHONE SWITCH
US4567590A (en) * 1983-12-27 1986-01-28 International Business Machines Corp. Message stripping protocol for a ring communication network
US4821174A (en) * 1984-03-20 1989-04-11 Westinghouse Electric Corp. Signal processing system including a bus control module
JPH0776932B2 (en) * 1984-12-07 1995-08-16 日本電気株式会社 Data transmission method
JPH0758967B2 (en) * 1986-02-10 1995-06-21 株式会社トキメック Fluid control valve communication control system
JPH0827738B2 (en) * 1986-08-15 1996-03-21 株式会社日立製作所 Online test method
US4835673A (en) * 1987-04-27 1989-05-30 Ncr Corporation Method and apparatus for sharing resources among multiple processing systems
US4965719A (en) * 1988-02-16 1990-10-23 International Business Machines Corporation Method for lock management, page coherency, and asynchronous writing of changed pages to shared external store in a distributed computing system
JP2633900B2 (en) * 1988-04-22 1997-07-23 株式会社日立製作所 Common bus control method
US5041963A (en) * 1988-12-29 1991-08-20 Intel Corporation Local area network with an active star topology comprising ring controllers having ring monitor logic function
CA2011934A1 (en) * 1989-07-19 1991-01-19 Theodore Heske, Iii Method and apparatus for source routing bridging
US5189246A (en) * 1989-09-28 1993-02-23 Csir Timing apparatus
US5247616A (en) * 1989-10-23 1993-09-21 International Business Machines Corporation Computer system having different communications facilities and data transfer processes between different computers
US4993025A (en) * 1989-11-21 1991-02-12 Picker International, Inc. High efficiency image data transfer network
US5265228A (en) * 1989-12-05 1993-11-23 Texas Instruments Incorporated Apparatus for transfer of data units between buses
US5261089A (en) * 1990-05-16 1993-11-09 International Business Machines Corporation Optimization of commit procedures by utilizing a two-phase commit procedure only when necessary
US5319773A (en) * 1990-05-16 1994-06-07 International Business Machines Corporation Asynchronous resynchronization of a commit procedure
US5165031A (en) * 1990-05-16 1992-11-17 International Business Machines Corporation Coordinated handling of error codes and information describing errors in a commit procedure
JP3293839B2 (en) * 1990-05-16 2002-06-17 インターナショナル・ビジネス・マシーンズ・コーポレーション Computer system that adjusts the commit scope to the unit of work
JP2691081B2 (en) * 1990-05-16 1997-12-17 インターナショナル・ビジネス・マシーンズ・コーポレイション Computer network
US5319774A (en) * 1990-05-16 1994-06-07 International Business Machines Corporation Recovery facility for incomplete sync points for distributed application
US5327532A (en) * 1990-05-16 1994-07-05 International Business Machines Corporation Coordinated sync point management of protected resources
US5276876A (en) * 1990-05-16 1994-01-04 International Business Machines Corporation Registration of resources for commit procedures
US5243596A (en) * 1992-03-18 1993-09-07 Fischer & Porter Company Network architecture suitable for multicasting and resource locking
US6101321A (en) * 1992-04-10 2000-08-08 Eastman Kodak Company Method and apparatus for broadcasting data in a ring connected multiprocessor
US9619841B2 (en) 1996-03-28 2017-04-11 Integrated Claims Systems, Llc Systems to assist in the creation, transmission, and processing of health insurance claims
US6003007A (en) * 1996-03-28 1999-12-14 Dirienzo; Andrew L. Attachment integrated claims system and operating method therefor
US5758161A (en) * 1996-05-24 1998-05-26 International Business Machines Corporation Testing method for checking the completion of asynchronous distributed collective operations
US5862340A (en) * 1996-05-24 1999-01-19 International Business Machines Corporation Method operating in each node of a computer system providing and utilizing special records for collective communication commands to increase work efficiency at each node
US5748959A (en) * 1996-05-24 1998-05-05 International Business Machines Corporation Method of conducting asynchronous distributed collective operations
US7099352B1 (en) * 2001-01-03 2006-08-29 Juniper Networks, Inc. System, apparatus, and method for increasing resiliency in communications
US7073001B1 (en) * 2002-04-03 2006-07-04 Applied Micro Circuits Corporation Fault-tolerant digital communications channel having synchronized unidirectional links
US7343418B2 (en) * 2002-06-03 2008-03-11 Microsoft Corporation Peer to peer network
CN1739265A (en) * 2003-10-17 2006-02-22 松下电器产业株式会社 Data transmission system, data transmitter, and transmitting method
US8688780B2 (en) 2005-09-30 2014-04-01 Rockwell Automation Technologies, Inc. Peer-to-peer exchange of data resources in a control system
US8307129B2 (en) * 2008-01-14 2012-11-06 International Business Machines Corporation Methods and computer program products for swapping synchronous replication secondaries from a subchannel set other than zero to subchannel set zero using dynamic I/O
US7761610B2 (en) * 2008-01-25 2010-07-20 International Business Machines Corporation Methods and computer program products for defining synchronous replication devices in a subchannel set other than subchannel set zero
US8516173B2 (en) * 2008-07-28 2013-08-20 International Business Machines Corporation Swapping PPRC secondaries from a subchannel set other than zero to subchannel set zero using control block field manipulation

Family Cites Families (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
GB1168476A (en) * 1966-05-17 1969-10-29 British Telecomm Res Ltd Improvements in or relating to data transmission systems
JPS52117004A (en) * 1976-03-29 1977-10-01 Hitachi Ltd Error control for loop transmission
US4103336A (en) * 1976-07-26 1978-07-25 International Business Machines Incorporated Method and apparatus for allocating bandwidth on a loop system coupling a cpu channel to bulk storage devices
GB1570923A (en) * 1977-02-14 1980-07-09 Gec General Signal Ltd Data communication apparatus
US4368512A (en) * 1978-06-30 1983-01-11 Motorola, Inc. Advanced data link controller having a plurality of multi-bit status registers
US4354229A (en) * 1980-03-10 1982-10-12 International Business Machines Corporation Loop initialization mechanism for a peer-to-peer communication system

Also Published As

Publication number Publication date
US4489379A (en) 1984-12-18

Similar Documents

Publication Publication Date Title
CA1180459A (en) Distributed data processing in ring-structured networks architected for full duplex peer-to-peer operation of processing stations and uninterruptible transfer of long data records between stations
RU2242092C2 (en) Method for resetting object of medium access control layer in broadband code-division multiple access communication system using high-speed downlink burst access
US5640392A (en) Signal receiving apparatus
KR100308719B1 (en) Method and device for using multiple fifo enhancing flow control and routing in communication receiver
EP0042447B1 (en) Flow control mechanism for block switching nodes
US5210749A (en) Configuration of srams as logical fifos for transmit and receive of packet data
EP0637416B1 (en) Method and apparatus for preserving packet sequencing in a packet transmission system
EP0525985B1 (en) High speed duplex data link interface
EP0459757A2 (en) Network adapter
US4642630A (en) Method and apparatus for bus contention resolution
JPH0578978B2 (en)
JPH0578979B2 (en)
US4930122A (en) Message transfer system and method
US5053946A (en) Token ring network having token request mechanism
JPS63279633A (en) Multi-address communication system
US4878218A (en) Data transmission method and data transmission system free from local traffic jam
EP0537297B1 (en) A method and a device for destination and source addressing in a packet network
US4698804A (en) Shared data transmission system
US4550401A (en) Delivery information packet switching system
KR970000667B1 (en) Communication control device
GB2188216A (en) Ring bus communications systems
JP2615305B2 (en) Ring network station
JPS59117846A (en) Data communicating equipment
JPH0974410A (en) Communication repeater
JPS59212053A (en) Data relay system

Legal Events

Date Code Title Description
MKEC Expiry (correction)
MKEX Expiry