Castout processing for duplexed cache structures

Details Castout processing for duplexed cache structures Castout processing for duplexed cache structures

1999-02-22

2002-08-20

Nguyen, Than (Department: 2187)

Electrical computers and digital processing systems: memory

Storage accessing and control

Hierarchical memories

C711S141000, C711S147000, C711S150000, C711S151000

Reexamination Certificate

active

06438654

ABSTRACT:

TECHNICAL FIELD
This invention relates in general to the field of data processing and, in particular, to the duplexing of cache structures located within a coupling facility of a computing environment.
BACKGROUND ART
A cache structure is a high-speed cache shared by one or more independently-operating computing units of a computing environment. In particular, cache structures are located within a remote facility, referred to as a coupling facility, that is coupled to the one or more independently-operating computing units. The computing units store and retrieve data from the cache structures.
Coupling facility cache structures can be configured in several different modes. of operation, one of which is a store-in mode. Store-in mode caches are used, for example, by the DB
2
database management facility of International Business Machines Corporation. A key attribute of the store-in mode is that changed data may be stored into the non-volatile memory of the coupling facility using the high performance coupling facility links. This avoids the delay in the execution of database transactions that result when the data is written to secondary storage (e.g., direct access storage devices (DASD)) using normal input/output (I/O) operations, and is an advantage of the coupling facility cache.
Subsystems who cache changed data in a coupling facility cache face a unique recovery/availability problem, which is not faced by those who either do not cache data or cache only unchanged data. For example, when a data item is modified and only written changed to the coupling facility cache structure, a subsequent failure of the coupling facility cache structure can cause the only existing current level of the data item to be lost. This results in a loss of data integrity. This loss of integrity window exists from the time the data item is written to the coupling facility cache until it is eventually castout to permanent storage, which may be a considerable time. At any given instant, a significant percentage of data stored in the coupling facility cache structure may be in this changed state, and thus vulnerable to loss should the coupling facility structure be lost.
To recover from such failures, subsystems have made use of recovery logs, which are hardened on permanent storage. Basically, during normal operation, as a given subsystem instance modifies a data item, it first writes a description of the data item update to its own recovery log along with a unique ordering indication (typically, a timestamp) showing when the update to the data was made relative to the other updates. Then, when the log update is complete, it writes the updated data item to the coupling facility cache structure. Given this, if the cache structure fails, a recovery process can reconstruct the most current version of the data by merging the recovery logs of all subsystem instances so that updates made by all instances can be observed; locating the most current copy of each data item in the log, using the ordering information associated with each of the logged updates; and writing the most current copy of each of the data items to permanent storage.
While the above approach allows the data to be recovered following the failure of a coupling facility cache structure, it is not an adequate solution for providing continuous availability of the shared data and of the coupling facility cache structure across such failures. The log merge and recovery update processing can take a long time, during which time the database is entirely unavailable for use by end users.
Thus, a need exists for a recovery technique that allows recovery from a failure with little or no perceived unavailability of the data to the end users. A further need exists for a mechanism that allows selected data to be duplexed. A yet further need exists for a mechanism that allows duplexing to be turned on and off automatically. A yet further need exists for a technique that enables a switch from duplex mode to simplex mode to be performed quickly.
SUMMARY OF THE INVENTION
The shortcomings of the prior art are overcome and additional advantages are provided through the provision of a castout method. In one embodiment, the castout method includes writing a selective data item from a primary instance of a data structure to at least one storage medium; determining whether the selective data item can be deleted from a secondary instance of the data structure; and deleting the selective data item from the secondary instance, when the selective data item can be deleted.
The duplexing capability of the present invention advantageously provides for improved availability of data, such as cache structure data. Duplexing can be initiated on a per-structure basis, either manually or automatically. Once duplexing is initiated, the operating system drives the structure users to temporarily quiesce access to the structure; allocate a secondary structure instance in, for example, a different coupling facility from the primary structure instance; copy any necessary structure data from the primary instance to the secondary instance, establishing a duplexed copy of the structure data; and unquiesce access to the structure with duplexing established.
Once duplexing is established, the user explicitly duplexes any necessary updates to both the primary and secondary structure instances to maintain synchronization.
When a structure failure or loss of connectivity affects one of the structure instances, the operating system drives the structure users to revert to simplex mode on the unaffected structure instance. The switch to simplex mode is very fast, with no data loss and no log recovery needed. Duplexing may then be reinitiated for the structure either automatically or manually.
At the time changed data in the cache structure is castout causing the data entries to be marked unchanged, the present invention advantageously deletes the entries from the secondary structure.
Additional features and advantages are realized through the techniques of the present invention. Other embodiments and aspects of the invention are described in detail herein and are considered a part of the claimed invention.


REFERENCES:
patent: 5280611 (1994-01-01), Mohan et al.
patent: 5305448 (1994-04-01), Insalaco et al.
patent: 5317739 (1994-05-01), Elko et al.
patent: 5339427 (1994-08-01), Elko et al.
patent: 5394542 (1995-02-01), Frey et al.
patent: 5455942 (1995-10-01), Mohan et al.
patent: 5457793 (1995-10-01), Elko et al.
patent: 5463736 (1995-10-01), Elko et al.
patent: 5465359 (1995-11-01), Allen et al.
patent: 5493668 (1996-02-01), Elko et al.
patent: 5515499 (1996-05-01), Allen et al.
patent: 5537574 (1996-07-01), Elko et al.
patent: 5561809 (1996-10-01), Elko et al.
patent: 5630050 (1997-05-01), Neuhard et al.
patent: 5634072 (1997-05-01), Allen et al.
patent: 5664155 (1997-09-01), Elko et al.
patent: 5706432 (1998-01-01), Elko et al.
patent: 5751992 (1998-05-01), Bhargava et al.
patent: 5813032 (1998-09-01), Bhargava et al.
patent: 5860115 (1999-01-01), Neuhard et al.
patent: 5875484 (1999-02-01), Neuhard et al.

Affiliated with

Also associated with

Rating

Castout processing for duplexed cache structures does not yet have a rating. At this time, there are no reviews or comments for this patent.

If you have personal experience with Castout processing for duplexed cache structures, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Castout processing for duplexed cache structures will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFUS-PAI-O-2938707