1 / 19

Replica Placement Algorithms for Mobile Transaction Systems

Replica Placement Algorithms for Mobile Transaction Systems. Manghui Tu, Peng Li, Liangliang XiaoI-Ling Yen, Farokh B. Bastani ,. Presented by: Tulasi Bobbala. Abstract. Major concerns in Distributed mobile systems: Communication cost and disconnections.

dugan
Download Presentation

Replica Placement Algorithms for Mobile Transaction Systems

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. Replica Placement Algorithms for Mobile Transaction Systems Manghui Tu, Peng Li, Liangliang XiaoI-Ling Yen, Farokh B. Bastani, Presented by: Tulasi Bobbala

  2. Abstract • Major concerns in Distributed mobile systems: • Communication cost and disconnections. • Replica replacement issues. • Solution - A heuristic “expansion-shrinking” algorithm is developed to make replica replacement decisions.

  3. Introduction • Many modern distributed applications support mobile clients to access backend databases and/or shared data files. • Due to the limited bandwidth and frequent disconnections in mobile environment, many efforts have been made to improve the performance and reliability for mobile transaction processing.

  4. Introduction – 2 • Here, the placement issues for partial replication of correlated data objects. • A data placement algorithm is proposed to dynamically replicate correlated data objects based on historical access patterns.

  5. System model and Problem Specification • PBC,PMC • DBC={d1,d2,d3,d4……dN} • DMC denotes the set of data objects replicated on mobile node PMC. • Transaction data set D(t) • They are two types of transactions 1) Read transaction 2) Update transaction

  6. Problem specification • To minimize the traffic on the link between PMC and PBC we need to determine the set of data objects to be placed on PMC. Where Q(s) is set of read transactions and W(s) is set of update Transactions.

  7. Replication Cost Model Based On Data Correlations Two cost models are needed for the data correlation model, including the replica allocation cost model and deallocation cost model. We define additional cost as: We define additional benefit as: We define access cost as:

  8. Replication cost model considering disconnections • The log LMC at PMC and LBC at PBC record the data sets of the update transactions and are being used by the replicaplacement algorithm to estimate the communication cost. • The modified costs are given in the following: Wa=Wb=1/the average number of update transactions in a group

  9. Optimal Partial Replication Algorithm Based on Data Correlations • OPR Algorithm • The OPR algorithm consists of optimal partial replica allocation (OPRA) executed at PBC and optimal partial replica deallocation (OPRD) executed at PMC. The algorithm is developed based on the depth-first branch and bound approach.

  10. The OPRA algorithm:

  11. OPRD The algorithm OPRD is developed in a similar way as OPRA. OPRD is executed at PMC, and it deallocates a data set to minimize the communication cost along the link between PMC and PBC.

  12. The OPRD algorithm:

  13. Heuristic Partial Replication Algorithm • The heuristic algorithm isdecomposed into heuristic replica allocation and deallocation algorithms. • Log reduction. • Heuristic Replica Allocation Algorithm • Set-Expansion heuristic replica allocation algorithm. • Set-Shrinking heuristic replica allocation algorithm.

  14. Set-Expansion heuristic replica allocation algorithm.

  15. Set-Shrinking heuristic replica allocation algorithm.

  16. Heuristic Replica Deallocation Algorithm • Deallocation algorithm is similar to allocation algorithm which uses set expansion algorithm • The deallocation algorithm tries to find minimal set of data objects S recorded in LMC such that • The deallocation algorithm by Set-Shrinking algorithm tries to find a minimal set of data objects s recorded such that

  17. Frequency-Based Partial Replication The scheme determines whether a data object should be allocated to PMC or deallocated from PMC by analyzing its access frequency We define additional cost as: We define additional benefit as:

  18. Conclusion The contributions include, 1. Developing cost models for accessing correlated data objects and for the effect of disconnection on communication cost. 2. Proving that the optimal replica placement problem for correlated data object model is NP, 3. Proposing the heuristic expansion-shrinking algorithm for the correlated data object model and show that it achieves near-optimal solutions, and 4. Proposing the weighted heuristic algorithm to take the disconnection effect into account and to achieve better data object placement.

  19. THANK YOU

More Related