• The Journal of Korean Association of Computer Education
• /
• v.5 no.2
• /
• pp.21-28
• /
• 2002

This paper proposes a new approach to rollback-recovery using multi-agent in distributed computing system. Previous rollback-recovery protocols are dependent on inherent communication and operating system, which causes a decline of computing performance in distributed computing system. By using multi-agent, we propose rollback-recovery protocol that is independent on operating system. We define three kinds of agent. One is a recovery agent that performs rollback-recovery protocol after a failure. Other is an information agent that constructs domain knowledge as a rule of fault tolerance and information during failure-free operation. The other is facilitator agent that controls the efficient communication between agents. Also we propose rollback-recovery protocol using multi-agent and simulated the proposed rollback-recovery protocol using JAVA and agent communication language in CORBA environment.

• The KIPS Transactions:PartA
• /
• v.13A no.5 s.102
• /
• pp.421-428
• /
• 2006

This paper proposes an integrated recovery approach applying retry and rollback techniques to recover the TMR failure. Combining the time redundancy techniques with W system is apparently effective to recover the TMR failure(or masked error) primarily caused by transient faults. These policies need fewer reconfigurations at the cost of extra time required for the time redundant schemes. The optimal numbers of retry and rollback to minimize the mean execution time of tasks are derived for the proposed method through computing the likelihoods of all possible states of the failed system. The effectiveness of the proposed method is validated through examining certain numerical examples and simulations conducted with a variety of parameters governing environmental characteristics.

• Journal of KIISE:Databases
• /
• v.27 no.4
• /
• pp.549-561
• /
• 2000

In database management systems(DBMSs), partial rollback is a useful facility that cancels part of the executed operations upon user's requests without a total rollback. Many relational DBMSs(RDBMSs) provide this facility, However, object-orientccd DBMSs (OODBMSs) cannot utilize the previous recovery scl18lne of partial rollback usccd in (RDBMSs) since, unlike RDBMSs, they use a dual buffer consisting of an object buffer and a page buffer. Therefore, a new recovery scheme is required that rolls back the data efficiently in the dual buffer. We propose four partial rollback schemes in OODBMSs that use a dual buffer. We classify the proposed schemes into the single buffer based partial rollback scheme and the dual buffer based partial rollback scheme according to the number of buffers used for partial rollback processing. We further classify them into Uthe page buffer based partial rollback scheme, 2)the object buffccr based partial rollback scheme, 3)the dual buffer based partial rollback scheme using soft log, and 4)the dual buffer based partial rollback scheme using shadows. We evaluate the performance by mathematical analysis and experiments. The results show that the dual buffer based partial rollback scheme using shadows provides the best performance.

• Journal of KIISE:Databases
• /
• v.30 no.1
• /
• pp.64-79
• /
• 2003

In client-server computing environments, update transactions of spatial data have the following characteristics. First, a transaction to update maps needs interactive work, and therefore it nay take a long time to finish. Second, a long transaction should be allowed to read the dirty data to enhance parallelism of executing concurrent transactions. when %he transaction is rolled back, it should guarantee the cascading rollback of all of the dependent transactions. Finally, two spatial objects may have a weak dependency constraint, called the spatial relationship, based on geometric topology. The existing recovery approaches cannot be directly applied to this environment, due to the high rollback cost and the overhead of cascading rollbacks. Furthermore, the previous approaches cannot guarantee the data integrity because the spatial relationship, which is a new consistency constraint of spatial data, is not considered. This paper presents new recovery schemes for update transactions of spatial data. To guarantee the data integrity, this paper defines recovery dependency as a rendition of cascading rollbacks. The partial-rollback is alto suggested to solve the problem of high rollback cost. The recovery schemes proposed in this paper can remove the unnecessary cascading rollbacks by using undo-delta, partial -redo and partial-undo. Finally, the schemes are performed to ensure the correctness.

• The Transactions of the Korea Information Processing Society
• /
• v.6 no.1
• /
• pp.216-224
• /
• 1999

In the paper we propose two recovery methods by adopting a rollback and/or roll-forward technique (S) to recover TMR failures in a TMR (structured ) system that is the simplest spatial redundancy. This technique is apparently effective to recovering TMR failures primarily caused by transient faults. The proposed policies carry out few reconfigurations at the cost of (minimal) time-overhead needed for those time-redundant schemes. The optimal checkpoint-interval vectors are derived for both methods through the likelihoods of all (possible) states of the system as well as the total execution-time. Consequently the effectiveness of our proposed policies is validated through certain numerical examples and simulations.

• Proceedings of the Korean Information Science Society Conference
• /
• 2002.10c
• /
• pp.403-405
• /
• 2002

본 논문은 Recovery 기법 중의 하나인 Roll-Forward를 적용한 고가용 웹 서버의 설계 및 구현에 대해서 그 방안을 제시한다. Roll-Forward Recovery 는 복구에 드는 시간을 감소시킬 수 있다는 장점이 있다. 즉, 일반적인 중복 시스템에서 결함이 발생하면 시스템의 상태는 가장 최근의 checkpoint 지점으로 Rollback 되어야 하기 때문에 복구에 많은 시간을 소비하게 되지만, Roll-Forward를 적용함으로써 복구에 드는 시간을 감소시킬 수 있다. 단일 서버시스템 또는 Rollback Recovery를 적용한 시스템의 결함 발생 시 서비스를 제공할 수 없는 상태가 되거나 복구에 많은 시간을 소비한다. 이에 비해 Roll-Forward를 적용한 중복 서버시스템을 구성하여 동작 운용중인 서버의 결함 발생 시 즉시 다른 서버가 결과를 보여줌으로써 고가용 웹 서비스를 제공할 수 있다. 본 논문에서 제시하는 Roll-Forward를 적용한 고가용 웹 서버의 동작원리는 마스터서버에 들어오는 요청을 백업서버에서도 똑같이 수행하며, 마스터 서버가 동작 중 결함이 발생하였을 경우 백업서버가 피 결과를 보여줌으로써 고가용의 서비스가 제공되는 웹 서버를 구현하고자 한다.

• Journal of KIISE:Databases
• /
• v.30 no.2
• /
• pp.197-208
• /
• 2003

Abstract Workflow is not appropriately applied to coordinated execution of applications(steps) that comprise business process such as a collaborative series of tasks because of the lacks of network infra, standard of information exchange and data consistency management with conflict mode of shared data. Particularly we have not mentioned the problem which can be occurred by shared data with conflict mode. In this paper, to handle data consistency in the process of rollback for failure handling or recovery policy, we have classified rollback dependency into three types such as implicit rollback dependency in a transactional workflow, implicit rollback dependency in collaborative transactional workflows and explicit rollback dependency in collaborative transactional workflows. Also, we have proposed the rollback dependency compiler that determines above three types of rollback dependency. A workflow designer specifies the workflow schema and the resources accessed by the steps from a global database of resources. The rollback dependency compiler generates the enhanced workflow schema with the rollback dependency specification. The run-time system interprets this specification and executes the rollback policy with data consistency if failure of steps is occurred. After all, this paper can offer better correctness and performance than state-of-the-art WFMSs.

• Journal of KIISE
• /
• v.42 no.1
• /
• pp.7-14
• /
• 2015

Storage Class Memory(SCM) has been considered as a next-generation storage device because it has positive advantages to be used both as a memory and storage. However, there are significant problems of data consistency in recently proposed file systems for SCM such as insufficient data consistency or excessive data consistency-control overhead. This paper proposes a novel data consistency-control scheme, which changes the write mode for log data depending on the modified data ratio in a block, using a rollback-recovery scheme instead of the Write Ahead Logging (WAL) scheme. The proposed scheme reduces the log data size and the synchronization cost for data consistency. In order to evaluate the proposed scheme, we implemented our scheme on a Linux 3.10.2-based system and measured its performance. The experimental results show that our scheme enhances the write throughput by 9 times on average when compared to the legacy data consistency control scheme.

• Journal of Institute of Control, Robotics and Systems
• /
• v.10 no.8
• /
• pp.748-754
• /
• 2004

In this paper, we propose the Triple Modular Redundancy (TMR) control system equipped with a checkpoint strategy. In this system, faults in a single processor are masked and faults in two or more processors are detected at each checkpoint time. When faults are detected, the rollback recovery is activated to recover from faults. The conventional TMR control system cannot overcome faults in two or more processors. The proposed system can effectively cope with correlated and independent faults in two or more processors. We develop a reliability model for this TMR control system under correlated and independent transient faults, and derive the reliability equation. Then we investigate the number of checkpoints that maximizes the reliability.

• The Journal of Korea Institute of Information, Electronics, and Communication Technology
• /
• v.14 no.3
• /
• pp.201-207
• /
• 2021

Deep learning, a type of machine learning, performs learning while changing the weights as it progresses through each learning process. Tensor Flow and Keras provide the results of the end of the learning in graph form. Thus, If an error occurs, the result must be discarded. Consequently, existing technologies provide a function to roll back learning results, but the rollback function is limited to results up to five times. Moreover, they applied the concept of MLOps to track the deep learning process, but no rollback capability is provided. In this paper, we construct a system that manages the intermediate value of the learning process by blockchain to record the intermediate learning process and can rollback in the event of an error. To perform the functions of blockchain, the deep learning process and the rollback of learning results are designed to work by writing Smart Contracts. Performance evaluation shows that, when evaluating the rollback function of the existing deep learning method, the proposed method has a 100% recovery rate, compared to the existing technique, which reduces the recovery rate after 6 times, down to 10% when 50 times. In addition, when using Smart Contract in Ethereum blockchain, it is confirmed that 1.57 million won is continuously consumed per block creation.