Dynamic magnetic information storage or retrieval – Automatic control of a recorder mechanism – Controlling the head
Reexamination Certificate
2002-01-29
2004-06-22
Hudspeth, David (Department: 2651)
Dynamic magnetic information storage or retrieval
Automatic control of a recorder mechanism
Controlling the head
Reexamination Certificate
active
06754036
ABSTRACT:
FIELD OF THE INVENTION
This application relates generally to determining a seek profile in a data storage device and more particularly to a system and/or method for automatically determining disc drive seek profiles for use in command queue ordering.
BACKGROUND OF THE INVENTION
Modern hard disc drives comprise one or more rigid discs that are coated with a magnetizable medium and mounted on a hub of a spindle motor for rotation at a constant high speed about a rotational axis. Information is stored on one or more surfaces of the disc or discs in a plurality of concentric circular tracks. An aggregate of tracks on the surfaces of a disc or discs at a given radial position from the rotational axis is referred to as a cylinder. Data is written to, and/or read from, sectors on the tracks via transducers (“heads”) mounted to a radial actuator, which positions the heads relative to the discs. The read/write elements are typically positioned over specific sectors of the disc in accordance with commands received from a host computer.
Typical disc drives use microprocessors to execute commands received from the host computer. Often, the disc drive will receive commands from the host faster than the commands can be processed by the microprocessor. When this occurs, the commands which are waiting to be executed are typically queued or cached in a “command queue” for later processing. As the commands coming into the disc drive are not necessarily received in an optimal order for processing, optimization of the queued commands is desirable. For example, a disc drive may receive commands to read and write data at a variety of locations on the hard discs within the disc drive. Optimally, these commands would be processed in a manner which would minimize the movement of the disc drives read/write heads across the disc. Ordering the commands in this manner is called command queue reordering. Command queue reordering allows for a more efficient use of the microprocessor as well as a more efficient use of the hardware being controlled by the microprocessor.
Traditionally, disc drives have sorted commands in an order that minimizes latency between the various commands. To determine if a seek can be completed within a given amount of latency, the seek time must be calculated. The seek time is the time required for the read/write element to radially move across or traverse cylinders between the present cylinder over which the read/write element is positioned and the cylinder to be addressed by the particular command. To determine the seek time for a particular command, disc drives often use what is referred to as a seek profile to define seek times versus seek distance in the disc drive. Often this seek profile is implemented as a seek time array containing entries for various seek times versus seek distance on the disc drive. This seek time array may be used on its own or, alternatively, a seek time array may be used as a single element or input into a more complex command queue reordering process that takes into account the rotational latency of the disc drive in determining the optimal order of commands in the command queue. An example of one such process is rotational position sorting (RPS). One rotational position sorting process is described in U.S. Pat. No. 5,570,332 for “Method for Reducing Rotational Latency in a Disc Drive” to Heath, et al., which is incorporated herein by this reference.
In its most precise form, a seek profile would indicate the precise seek time for each seek distance that may be traversed in a disc drive. For example, this type of seek profile could comprise a seek time array containing an individual entry for each possible seek distance on a disc. Unfortunately, as modern disc drives contain tens of thousands of tracks on a single disc, the memory requirements for storing and maintaining a comprehensive seek time array of this type is prohibitive. One approach that has been used to reduce the memory requirements of seek profiles involves using a partial seek time array to compute seek times for relatively short seeks and a simple linear equation to compute seek times for relatively longer seek distances. This approach takes advantage of a particular characteristic of the overall seek profile of many, particularly older, disc drives. That is, the relationship between seek times and seek distances in the portion of the seek profile corresponding to relatively longer seek times is substantially linear. As such, the seek profile in this so called “linear region” can be characterized using a linear equation that, in turn, is used to predict a seek time for a given seek distance. In contrast, the relationship between seek times and seek distances in the portion of the seek profile corresponding to relatively shorter seek times is substantially non-linear. As such, a partial seek time array is used to compute seek times in this “non-linear” region of the seek profile, where a simple linear equation would not accurately model the seek profile.
The seek time array, as described so far, may be either static or dynamic. A static seek time array is a precomputed look-up table constructed from a static seek profile. The static seek profile consists of piecewise-linear approximations of actual seek performance. These approximations are typically determined manually by a test engineer after viewing a graph or some other representation of actual seek times versus seek distances as measured in a representative disc drive of a particular model type. The piecewise-linear approximation typically comprises a number of straight line segments, each assigned to a non-overlapping range of seek lengths, which together estimate performance for seeks within the non-linear range. This piecewise-linear approximation is then used to construct the look-up table of the static seek time array, which is stored in memory on the disc drive for use during operation.
While the use of a static seek time array in the manner just described has proven relatively effective, there are still a number of problems associated therewith. One such problem relates to the manner in which straight line segments for the piecewise-linear approximations defining the static seek time array are determined. As mentioned, these segments are typically defined manually as approximations of the actual seek performance by a test engineer after viewing seek data taken from a representative disc drive. As such, the accuracy of these approximations is limited by the skill, or the lack thereof, of the test engineer who makes these approximations. Another problem associated with the way static seek arrays of this type array are determined is that the seek characteristics of a given model of disc drive may vary significantly from drive to drive. As such, a static seek time array that is computed based on the results of one representative drive of a particular model, may not be accurate for other drives of the same model. Additionally, seek characteristics of a single disc drive may change measurably over time, thus degrading the accuracy of the static seek array.
An alternative to the static seek time array, and one that addresses some of the noted problems of the static seek array, is a dynamic or adaptive seek time array. Similar to the static seek array, the adaptive seek array is an array or look-up table containing estimated seek times required for read/write element to traverse a given seek distance or length, in cylinders. The adaptive seek array typically uses the static seek array to initialize its look-up table. However, each time read/write element traverses a given number of cylinders to execute a command, the system or firmware of the disc drive senses the actual seek time necessary for the read/write element to traverse the given seek distance. Based on the sensed actual seek times for seek distances, the estimated seek times corresponding to seek distances within the adaptive seek array are modified.
While an adaptive seek array addresses some of the problems of the static seek array, adaptive seek arrays still retain so
Berger Derek J.
Hudspeth David
Lucente David K.
Seagate Technology LLC
Slavitt Mitchell
LandOfFree
Automated tuning of disc drive seek profile does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Automated tuning of disc drive seek profile, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Automated tuning of disc drive seek profile will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3318329