Electrical computers and digital data processing systems: input/ – Input/output data processing – Flow controlling
Reexamination Certificate
2000-06-28
2003-10-21
Gaffin, Jeffrey (Department: 2182)
Electrical computers and digital data processing systems: input/
Input/output data processing
Flow controlling
C710S001000, C710S020000, C710S031000, C709S233000, C709S250000, C711S004000
Reexamination Certificate
active
06636908
ABSTRACT:
BACKGROUND OF THE INVENTION
The present invention relates to data management. More specifically, the present invention relates to methods and apparatuses for managing large amounts of data, for example, in storage area networks and mainframe I/O environments.
The demands for managing large amounts of data have steadily increased in recent years and are expected to continue to increase in the future. For example, large organizations such as airlines and financial institutions require continuous, reliable, around-the-clock access to their “mission critical” data. Temporary interruptions in the accessibility to this data, or the loss of portions of this data, can be catastrophic to such organizations. Complicating the management task, many organizations have an enormous and growing amount of mission critical data (e.g., many terabytes). Much of this data is managed by mainframe-based computer systems.
FIG. 1
shows a block diagram of an exemplary prior art computing system
100
, which is described here to illustrate common management tasks and associated problems. System
100
includes two mainframe computers
110
A,
110
B, three direct access storage devices (DASDs, also known as “control units”)
120
A,
120
B,
120
C, and a director
150
. The various components communicate with one another via “point-to-point” communication links
160
,
162
,
164
,
166
,
168
,
170
, and
172
according to a defined protocol. The common protocol is the ESCON protocol, also known as the SBCON protocol (hereinafter, collectively called “ESCON”).
In this exemplary system, port
114
A of computer
110
A is coupled to port
122
B of control unit
120
B via link
162
. Port
112
B of computer
110
B is coupled to port
124
B of control unit
120
B via link
164
. Port
114
B of computer
110
B is coupled to a port (not shown) of director
150
via link
166
. Port
126
B of control unit
120
B is coupled to another port (not shown) of director
150
via link
168
. Port
124
C of control unit
120
C is coupled to a port (not shown) of director
150
via link
170
. In each case, the physical link and protocol are ESCON compliant. Finally, port P
1
of control unit
120
A is coupled to port P
2
of control unit
120
B via a proprietary communication link
172
, in which the physical link is ESCON but which is used to carry proprietary commands and data, for example, to implement certain “extended functions” (more below). For convenience of illustration, each computer
110
A,
110
B is illustrated as including two ESCON ports
112
,
114
.
In the illustrated embodiment, the mainframe computers are IBM S/390s. Exemplary IBM S/390 mainframe computer may include between sixteen to 256 ESCON communication ports.
Each exemplary DASD control unit
120
, for ease of description, is shown as including three ESCON communication ports
122
,
124
,
126
, and optionally additional private links P
1
, P
2
, though a typical DASD control unit may include between 2 and 64 ESCON ports. The illustrative DASD control units
120
include a main memory
130
, a controller
132
, a persistent storage
134
, and three memory buffers
123
,
125
,
127
, each buffer being associated with a corresponding ESCON port. Each port can write data into, and read data out of, its associated buffer. The controller
132
can write data into, and read data out of, all of the buffers or move data to or from persistent storage
134
via an internal system bus
136
.
A director
150
improves connectivity in a storage network by allowing one mainframe computer port to connect to two or more control units.
As mentioned above, the various components may communicate using the ESCON protocol. Under ESCON, the components communicate according to “chains” of one or more channel command words (CCWs). Each CCW, in turn, is communicated in three phases: a “command phase,” a “data phase,” and a “status phase” with each phase using a known vocabulary of messages. During each phase, information is transmitted as “frames,” which are 1 kilobyte or less in size and include control (or header) and data (or payload) portions. A given phase may involve known flow control and/or handshaking and may involve many frames. For example, the protocol permits 64 kilobyte transfers, which could require 64 frames during the data phase. The data phase uses a flow control technique in which an initiator expresses a desire to transmit or read a certain amount of data (e.g., in a prior write command), and the receiver replies with a data request message indicating the size of data that may be sent by the transmitter and received by the receiver. A series of such requests may be needed to transfer the entire “exchange.”
The I/O protocols rely on a concept of virtual links connecting “virtual mainframe machines” with “virtual control units. Some of the I/O protocols, which are connection oriented, like ESCON and SCSI, allow only one connection to be active at any moment in time, while others may actually frame multiplex the information among the various virtual links. Virtual links are effectively identified by the frame header information specifying both physical and logical addresses, and the components can detect virtual connections and disconnections from analyzing specific bits in certain frames.
As alluded to above, commercially available control units offer “extended functions.”Extended functions implement features above and beyond basic device operations like read or write. (The actual functions implemented by a device are defined in the device specification, such as a specification of a control unit.) For example, two popular extended functions are known as “concurrent copy” and “remote copy,” which are used, respectively, for maintaining backup copies or for “mirroring” data to other storage as it is written to its target. Known extended functions operate at a physical level of addressing (e.g., volume numbers and tracks) as opposed to operating at the logical level (e.g., files or the like). Referring back to
FIG. 1
, a control unit may perform back-up to another disk controlled by another control unit by using a dedicated ESCON link
172
, connecting the two control units. Proprietary software (sometimes referred to as firmware), executing on the control units, performs the necessary operations over the link
172
to send the data to be backed up from one control unit to another.
In the above approach to mirroring, data is effectively written to the control units sequentially, first to the primary control unit and then from the primary control unit to the control unit doing the mirroring. This introduces delay and complication as the data is written between the control units. The backup approach is also sequential. These approaches require dedicated communication links
172
that cost port connections on the control units.
Moreover, because prior art extended functions are built using proprietary embedded software (also known as “firmware”) to and between control units, third parties cannot practically create additional functions for the control units. To date, the extended functionality is largely limited to homogenous systems of control units. That is, the extended functions generally do not work when control units from different manufacturers are involved in a network.
Clustering is similar to mirroring in that some data is effectively mirrored to storage associated with another processor. However, rather than mirroring information in case a subsequent failover or switchover to another storage proves necessary, clustering usually involves mirroring (or replicating ) only specific information so that the processors may act collaboratively and in distributed fashion.
SUMMARY OF THE INVENTION
The inventions provides devices, systems, and methods of replicating and manipulating I/O information to improve efficiency and functionality. Preferably, the invention intercepts I/O information as it is transmitted between a computer (e.g., mainframe) and storage system (e.g., DASD storage controller).
Under certain aspects of the invention, an intelligent sp
Mokryn Marek
Mokryn Seweryn
Winokur Alexander
Casiano Angel L
Gaffin Jeffrey
Hale and Dorr LLP
SANgate Systems, Inc.
LandOfFree
I/O system supporting extended functions and method therefor does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with I/O system supporting extended functions and method therefor, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and I/O system supporting extended functions and method therefor will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3139089