Multiplex communications – Diagnostic testing
Reexamination Certificate
2000-11-22
2002-12-03
Yao, Kwang Bin (Department: 2664)
Multiplex communications
Diagnostic testing
Reexamination Certificate
active
06490253
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates to the field of loop diagnostics. More particularly, this invention relates to a peer-to-peer interface diagnostics.
BACKGROUND OF THE INVENTION
One key component of any computer system is a device to store data. Computer systems have many different places where data can be stored. One common place for storing massive amounts of data in a computer system is on a disc drive. The most basic parts of a disc drive are a disc that is rotated, an actuator that moves a transducer to various locations over the disc, and electrical circuitry that is used to write and read data to and from the disc. The disc drive also includes circuitry for encoding data so that it can be successfully retrieved and written to the disc surface. A microprocessor controls most of the operations of the disc drive as well as passing the data back to the requesting computer and taking data from a requesting computer for storing to the disc.
Information representative of data is stored on the surface of the storage disc. Disc drive systems read and write information stored on tracks on storage discs.
Fibre Channel (FC) is a serial data transfer architecture standardized by ANSI. A prominent FC standard is Fibre Channel Arbitrated Loop (FC-AL). This standard defines a distributed daisy-chained loop. FC provides for peer-to-peer communication on this loop.
FC-AL was designed for new mass storage devices and other peripheral devices that require very high bandwidth. FC-AL supports the Small Computer System Interface (SCSI) command set in addition to other upper level protocols. The mapping of these upper level protocols to FC is referred to as the FC-4 layer.
In a FC-AL, information from an originating device can pass through multiple other devices, and the links between the devices, before arriving at the recipient device. While the passage of information on multiple links adds complexity to isolating marginal and failing links over point to point connections, three conventional techniques of isolating marginal links exist. One technique of isolating marginal FC links uses link status to isolate the problem link. A second approach uses error-reporting features of the FC-4 mapping. A third approach is a combination of the first two.
A primary requirement for the three techniques is knowledge of the topology (i.e. connection order). Knowledge of the topology may be obtained during FC-AL defined loop initialization from the loop position map or by implicit means. An example of an implicit means is an enclosure of disks drives using hard addresses.
A first approach of using link status in the isolation of marginal links requires a management application (MA) in at least one of the devices on the loop. Several MAs may be implemented to cover the failure of any one. An MA may either periodically poll the loop during normal loop operation or request devices detecting link errors to report the incident. In polling mode, the link status accumulated in all the devices is used to locate marginal links. In the report error, identification mode, status accumulated from all devices reporting errors is used to locate marginal links.
The isolation of the source of a single error is possible with this approach but not guaranteed.
The use of link status makes the approach FC-4 independent. This is an advantage in multiple protocol loops. However, the drawback of using link status is the polling or report error mode overhead reduces the efficiency of the loop.
The second approach uses error-reporting features of the FC-4 mapping. Using FC-4 reported errors to isolate the source of errors on loops requires maintaining a log of the errors. The source is located by analyzing the log to determine which devices are reporting errors and which are not.
Using FC-4 reported errors to isolate the source of errors on loops removes the requirement for an MA to maintain link error history and poll the loop. Not polling the loop reduces overhead on the loop. Additionally, errors are only report when they occur.
Using FC-4 reported errors to isolate the source of errors on loops performs best in implementations in which a single master device receives all the reported errors. An example of such an implementation is a single initiator SCSI storage subsystem.
There are at least three drawbacks to relying on just the FC-4 error status. The occurrence of a single error does not provide sufficient information to isolate the source. Furthermore, status must be accumulated to build a history in order to isolate the error source. Lastly, in loops supporting multiple protocols or multiple devices that receive FC-4 status, implementation becomes difficult because the errors are not reported to a common destination device.
The third technique of isolating marginal FC links uses link status and FC-4 error reporting to isolate the problem link. Polling is not used and the isolation of the source of a single error is possible.
As with any use of link status, an MA is needed to keep error counts of all devices. When a FC-4 error is reported or the MA detects a link error, the MA reads the accumulated link status from all devices to determine the possible source of the error.
A disadvantage to implementation on loops with multiple FC-4s is the MA must support all the FC-4s.
Referring to
FIG. 1
, a diagram of a loop
100
comprised of SCSI Fibre Channel Protocol (FCP) devices, is shown. The loop includes a SCSI initiator device,
110
, that serves as the loop master, communicating with SCSI target devices
120
,
130
, and
140
. The link or interconnect
150
between devices
120
and device
130
is marginal and/or failing.
Error detection and reporting provided by a FC-4 may be used for the isolation of marginal links when available.
Due to the marginal link
150
, loop master
110
will experience command time-outs and data errors. The command time-outs are the result of errors during command, transfer ready, or response frames. These frames are discarded when they are received in error. Because the time-out could result from discarded frames to the targets, commands, or from the targets, transfer readies and responses, the location of the bad link can not be determined.
On write data operations, device
120
does not experience errors on data from loop master
110
. Device
120
and device
130
will, however, detect the errors introduced by the marginal link. Errors on write data are reported in the FCP Response.
On read data operations, loop master
110
does not detect errors on read data from device
130
and device
140
.
What is needed are loop error diagnostics that do not require knowledge of the topology of the loop, that reduces loop overhead traffic, and increases the effectiveness of the diagnostics.
SUMMARY OF THE INVENTION
In a peer to peer approach to isolation of error source, the management application (MA) function is distributed to all devices on the loop. Link status is used for error source isolation. More specifically, each device maintains the identity and the link error status of the device connected to its input, upstream device. When a device detects a link error on its input, the device initiates a request to the upstream device for link error counts.
When the link status for the upstream device indicates that device is also detecting link errors, the source of the errors is a different link on the loop. If the link status from the upstream devices does not indicate it is detecting errors, the source of the errors is likely the interconnect between the upstream device and the device itself. The device may then initiate diagnostic transfers between the upstream device and itself to verify the interconnect is marginal.
Advantageously, the present invention of loop error diagnostics does not require knowledge of the complete topology of the loop. The present invention also reduces loop overhead traffic because error isolation is distributed to each of the devices in the loop. Furthermore the effectiveness of loop diagnostics is increased because devices closest to the source of
Coomes James Allen
Miller Michael Howard
McCarthy Mitchell K.
Seagate Technology LLC
LandOfFree
Peer to peer interconnect diagnostics does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Peer to peer interconnect diagnostics, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Peer to peer interconnect diagnostics will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2985341