Error detection/correction and fault detection/recovery – Data processing system error or fault handling – Reliability and availability
Reexamination Certificate
1999-11-17
2003-09-30
Baderman, Scott (Department: 2184)
Error detection/correction and fault detection/recovery
Data processing system error or fault handling
Reliability and availability
C714S047300
Reexamination Certificate
active
06629266
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to software rejuvenation, and more particularly to transparent symptom-based selective software rejuvenation.
2. Description of the Related Art
The software executing in computer and networking systems can exhibit a behavior such that its failure rate increases over time, typically because of programming errors that generate increasing and unbounded resource consumption, or due to data corruption and numerical error accumulation (e.g., round-off errors). Examples of the effects of such errors are memory leaks, file systems that fill up over time, and spawned threads or processes that are never terminated.
The above-mentioned effects constitute a phenomenon known as “software aging,” and may be caused by errors in either application, middleware, or operating system software. As the allocation of a system's resources gradually approaches a critical level, the probability that the system will suffer an outage increases. This may be viewed as an increase in the software system's failure rate. Such a software system failure may result in overall system failure, “crashing”, “hanging”, performance degradation, etc.
One way of reducing the software's failure rate is to reset a portion of the system to recover any lost and unused resources. For example, this may be resetting just the application that is responsible for the aging, or it may be resetting the entire computer system. This is referred to as “software rejuvenation.” When only a part of the system is selected for rejuvenation, this is called “selective rejuvenation.”
FIG. 1
is a diagram showing the relationship of the software failure rate over time. As shown software ages over time as shown by reference numeral
100
, and the effects of software rejuvenations are shown at reference numeral
110
.
When the part of the system that is undergoing aging is reinitialized via rejuvenation, its failure rate falls back to its initial (e.g., lower), level because resources have been freed up and/or the effects of numerical errors have been removed. This has a dramatic effect on overall system availability. However, when the failure rate begins to climb again due to the above-mentioned causes, subsequent rejuvenations become necessary.
The transparent nature of this function is partly achieved by being incorporated into a management framework separate and independent of the operating system and application. This approach not only assists in providing transparency, but also allows for portability and interoperability across multiple operating systems. This is important as it is becoming more common for systems management to span multiple platforms in the enterprise.
It should be noted that error-induced resource exhaustion is a fundamentally different phenomenon from performance-based capacity exhaustion. “Error-induced resource exhaustion” is due to the causes outlined above and its effects can be eliminated temporarily by rejuvenation. “Performance-based capacity exhaustion” results from increases in performance requirements over time, and its effects can only be eliminated by reducing the computational performance requirements or adding new resources to the system.
In a conventional method and system, a time-based rejuvenation is performed within a single node and a multi-node environment. However, the conventional method and system assumes that the application must be modified to include calls to a fault tolerance library. Thus, modification of the application software source code is required. This is problematic because if the application source or executable code must be modified, then software rejuvenation technology cannot be applied to the broad class of applications for which the source or executable code cannot be economically modified, such as “shrink-wrapped” applications. Transparency, e.g., the lack of necessity to modify the application source or executable code, allows software rejuvenation to be applied to any application, regardless of whether the code can be modified or not. This extends its applicability to essentially all applications that can be executed on the computer system.
Additionally, the conventional method typically requires a proprietary dedicated system support for the rejuvenation functionality. Thus, the conventional methods and systems significantly restrict the rejuvenation's potential applicability and ease of use to systems that support the proprietary rejuvenation functionality.
In the above-mentioned co-pending application, it was described how to periodically rejuvenate all or part of a software system to reduce its failure rate to its initial, lower level, based on time.
However, in the co-pending application, there is no mechanism to notify the system operator to only rejuvenate prior to an unplanned outage, as opposed to at statically predetermined intervals.
Further, some system outages are characterized by not just one parameter reaching an extreme value, but by several values reaching a characteristic region in which not all values may be global extreme. In these cases, single-parameter extremes monitoring is ineffective.
Further, hitherto the invention, there has been a problem of predicting impending resource exhaustion. Some methods monitor a small number of key resource parameters of the operating system (e.g., the UNIX operating system), performing trend detection and slope estimation for those parameters, and projecting the mean time to their exhaustion based on the extrapolated trend. However, this technique has been limited to confirming the existence of software aging due to resource exhaustion. In this approach, there has been no successful prediction of outage since only individual indicators of outage are examined independently. There has been no approach of examining several indicators in combination to form a predictor of outage, even though none is necessarily at a global extreme.
Finally, the conventional methods have been unable to pinpoint and identify which subsystem, process, and thread are causing the exhaustion, and selectively rejuvenate solely the offending function.
SUMMARY OF THE INVENTION
In view of the foregoing and other problems, disadvantages, and drawbacks of the conventional methods and structures, an object of the present invention is to provide a method and system for predicting outages of a software system.
Another object is to provide a method and system for providing non-intrusive symptom-based prediction (and subsequent rejuvenation) of unplanned software outages.
In a first aspect of the invention, a method (and system) for increased software dependability, includes learning how to predict an outage of a software system running on a computer, based on the learning, predicting an imminent outage, and avoiding the outage.
In a second aspect of the invention, a method for software rejuvenation, includes waiting for symptoms associated with an outage of the software, determining whether a fail-to node has adequate resources to accept a failover workload, if the determining determines that the fail-to node cannot accept the failover workload, sending an alert that adequate redundancy does not exist to support fault tolerance requirements, and suspending rejuvenation until an operator acknowledges and corrects the deficiency.
In another aspect of the invention, a signal-bearing medium (e.g., storage medium) is provided for storing a program representing the inventive method.
Thus, the present invention provides symptom-based rejuvenation in several environments (e.g., the preferred embodiment being within a clustered environment). In the invention, the application need not be modified or to include calls to a fault tolerance library. Indeed, no such modification is necessary, and applications obtain the benefits of software rejuvenation without any modification. In addition, no proprietary, dedicated system support is required for the rejuvenation functionality. The use of rejuvenation by the invention is performed within the context of any industry-
Harper Richard Edwin
Hunter Steven Wade
Pahel, Jr. Thomas Dixon
Vaidyanathan Kalyanaraman
Zeggert William Paul
Baderman Scott
Lohn Joshua
McGinn & Gibb PLLC
Zarick, Esq. Gail H.
LandOfFree
Method and system for transparent symptom-based selective... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method and system for transparent symptom-based selective..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and system for transparent symptom-based selective... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3019504