Electrical computers and digital processing systems: memory – Storage accessing and control – Specific memory composition
Reexamination Certificate
1999-12-13
2004-03-30
Ellis, Kevin L. (Department: 2188)
Electrical computers and digital processing systems: memory
Storage accessing and control
Specific memory composition
Reexamination Certificate
active
06715034
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
This invention relates to routing or switching file system requests in a mass storage system, such as one having multiple storage devices.
2. Related Art
Mass storage systems can include multiple storage devices, such as disk drives. When a file server request arrives at the mass storage system, the request is identified as destined for one of the multiple storage devices, and is sent to that identified storage device for processing (and for a possible response). In a first set of known devices, the file server request is received by a hub device, which propagates the file server request to its target storage device, using a fiber channel arbitrated loop interface. Thus, the hub is coupled to each shelf of mass storage devices (where each shelf includes an enclosure housing a plurality of storage devices) using a star configuration of fiber channel arbitrated loop; thus, the hub couples each shelf to the loop and thus couples each of the mass storage devices to the loop. In a second set of known devices, the file server request is received by a switching device, which identifies the target storage device within the mass storage system. The switching device then sends the file server request directly to the shelf of mass storage devices including the target storage device, using a fiber channel arbitrated loop interface (the loop including each of the mass storage devices in that shelf); the switching device receives responses directly from that shelf of mass storage devices and routes those responses back to the original requesting device making the file server request.
One problem with known systems is that the fiber channel arbitrated loop is subject to error, including the possibility that the fiber channel has a break, disconnection, or other failure, and the additional possibility that one or more of the storage devices in the mass storage system will fail to forward file system requests to subsequent storage devices in the fiber channel arbitrated loop. In known devices including a hub, the hub monitors each connection to each shelf of mass storage devices (thus, paying attention to each of its ports, as well as to timing and signal requirements for signals on each connection). If the hub detects an error on a loop to or from a mass storage device in a shelf of storage devices, it bypasses the connection to that one shelf of storage devices, and effectively deletes the connectivity to all storage devices in that one shelf.
Although these known systems generally achieve a result of isolating errors in the fiber channel arbitrated loop, they are subject to several drawbacks. In cases when the hub detects errors, file server requests can fail to be sent to the target storage device. Similarly, in cases when the hub detects errors, even if the file server request is properly sent to the target storage device, one or more responses from the target storage device can fail to be properly sent back to the original requesting device. In systems where a hub or switch is used to route file server requests among multiple shelves of mass storage devices, isolating one of the fiber channel arbitrated loops has the effect of isolating an entire shelf of mass storage devices; in these cases, advantages obtained from redundancy of mass storage devices in the shelf (such as in RAID—Redundant Array of Inexpensive Disk—systems) are generally lost.
A first known method for attempting to remedy this problem is to provide redundant system elements, such as a plurality of fiber channel loops, a plurality of couplings between fiber channel loops and individual storage devices, or a plurality of elements at each other single point of failure. Although this first known method generally achieves the result of guarding against failure of the mass storage system, it is subject to several drawbacks. First, redundant system elements have additional cost, thus increasing the cost of the entire mass storage system. Second, system design using redundant system elements is more complex than without using redundant system elements, again increasing the cost of the entire mass storage system. Third, fiber channel drives are subject to failure modes that can bring down multiple fiber channel loops at once (for example, a bad crystal in a fiber channel drive will cause a mismatch in frequency and bring down both fiber channel links to which the fiber channel drive is coupled).
A second known method for attending to remedy this problem is to use serial channel techniques other than fiber channel arbitrated loops. One such alternative serial channel technique is SSA (Serial Storage Architecture), which is less subject to the drawbacks noted of the known art. However, use of SSA is subject to other drawbacks—fiber channel arbitrated loop is a known standard, and thus makes it easier to design systems for use with that known standard.
Accordingly, it would be advantageous to provide a technique for routing or switching file system requests in a mass storage system that is not subject to drawbacks of the known art. Such a technique would preferably include a capability for routing or switching file system requests in a mass storage system that is not subject to failure of either a fiber channel arbitrated loop or of any individual storage device.
SUMMARY OF THE INVENTION
The invention provides a method and system for routing or switching requests in a mass storage system. In a preferred embodiment, the mass storage system includes multiple storage devices coupled using fiber channel to a routing or switching device disposed internally within each shelf of mass storage devices in the mass storage system, and coupled directly to each individual mass storage device in that shelf. The switching device receives requests, identifies an individual target storage device for each such request, and sends each such request to its designated target storage device. The switching device also receives responses to such requests from each individual storage device within the shelf, and sends those responses from the storage device to an external medium with which the response can be delivered to the original requester. One advantage of the switching device is that the switching device can maintain communication bandwidth between the mass storage devices and external devices approximately equal to the communication bandwidth of a single fiber channel times the number of point-to-point connections.
The invention provides an enabling technology for a wide variety of applications for routing or switching requests in a mass storage system, so as to obtain substantial advantages and capabilities that are novel and non-obvious in view of the known art. Examples described below primarily relate to mass storage systems including a plurality of storage devices coupled using fiber channel arbitrated loop, but the invention is broadly applicable to many different types of storage systems.
REFERENCES:
patent: 4825354 (1989-04-01), Agrawal et al.
patent: 4937763 (1990-06-01), Mott
patent: 4984272 (1991-01-01), McIlroy et al.
patent: 5067099 (1991-11-01), McCown et al.
patent: 5113442 (1992-05-01), Moir
patent: 5144659 (1992-09-01), Jones
patent: 5202983 (1993-04-01), Orita et al.
patent: 5222217 (1993-06-01), Blount et al.
patent: 5257391 (1993-10-01), DuLac et al.
patent: 5283830 (1994-02-01), Hinsley et al.
patent: 5572711 (1996-11-01), Hirsch et al.
patent: 5617568 (1997-04-01), Ault et al.
patent: 5668958 (1997-09-01), Bendert et al.
patent: 5675782 (1997-10-01), Montague et al.
patent: 5689701 (1997-11-01), Ault et al.
patent: 5737523 (1998-04-01), Callaghan et al.
patent: 5761669 (1998-06-01), Montague et al.
patent: 5819292 (1998-10-01), Hitz et al.
patent: 5825877 (1998-10-01), Dan et al.
patent: 5841997 (1998-11-01), Bleiweiss et al.
patent: 5854903 (1998-12-01), Morrison et al.
patent: 5876278 (1999-03-01), Cheng
patent: 5890959 (1999-04-01), Pettit et al.
patent: 5915087 (1999-06-01), Hammond et al.
patent: 5931935 (1999-08-01), Cabrera et al.
patent: 5963962 (1999-10-01), Hitz
Aiello Anthony
Aster Radek
Kleiman Steven R.
Ellis Kevin L.
Network Appliance Inc.
Swernofsky Law Group PC
LandOfFree
Switching file system request in a mass storage system does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Switching file system request in a mass storage system, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Switching file system request in a mass storage system will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3278620