Effects of Semantic Cache on Mobile Database System

Transcription

1 Effects of Semantic Cache on Mobile Database System S.QADRI, A.R SATTAR & T. MUSTAFA Department of Computer Sciences University of Agriculture Faisalabad PAKISTAN Abstract: - In last few years database accessibility is a great issue especially in the distributed database environment. Recently mobile computing is becoming a hot issue and this research work is focused on some important features of semantic consistency techniques with implementation strategies. Researchers are trying to find out the basic factors which affect the best performance of semantic cache implementation strategies. It is also tried to find out the methodology how these issues can be solved in the best and efficient way. This research deals with the comparative performance study of inter/intra file semantic and fully associative techniques in detail and also suggested the best solution for analyzing these Software (simulator) had been used for important calculation which generated comparative results. For statistical data analysis SPSS software is used which provide the comparison between these Key-Words: - Semantic cache, Mobile Database, LRU, Simulator, Deadbase 1 Introduction Database has many issues to work in distributed environment especially in mobile databases. It is a hot issue because now days mobiles are very common but their reliability and accessibility is an important issue. The question is raised how, why and where anyone can access the entire data. It is the basic concept of mobile database. Similarly on the run time when and where you have to access of complete information is called mobile computing. This is all due to the advancement in network technology. Modern technology plays a vital role in the field of mobile computing. With out fast communication database is called deadbase because it provides or stimulates the information exchange from one point to another point very quickly. For mobile database it is important to see the factors which effects on performance. There are many factors which focused on performance like cache size which affects its performance. Greater the cache size, better the performance because more data comes in cache, similarly speed will be fast to access CPU Data. So available storage in cache also affects system performance. Replacement algorithms for data access play a vital role in mobile database performance. This research is emphasized on the semantic cache techniques (inter/intra file) with comparison of fully associative technique for best performance in mobile database. 2 Material & Methods 2.1 Semantic-based Caching Among different file access techniques, semantic based cache is different which uses the inherent relationship between (inter/intra files) to perform the functioning or hoarding [1].The proposed approach, does not use this relationship to prefetch files or save them, rather it focused on its efforts on saving these relationships in a cache. An eviction or removal index was calculated based on these relations. In this approach strong relations were saved but weak were omitted. This approach not only gave effective hit ratios but also decreases the communication overhead as compared to other replacement algorithms such as LRU, LFU and Greedy-dual size [2] when run against the distributed file system [3] and also on different application conditions. This approach was thought to be important as in semantic cache the relationships between files are taken that conventional approaches are mostly rejected or ignore. In semantic cache it minimizes the need for replacement algorithm which could mean increased file availability and reliability and

2 reduced synchronization overhead. User patience depends on importance of the used files and expected fetch delay [4]. Importance of a priority files are specified by user in some systems [4]. In the present study a thought had been developed to remove this overhead by establishing files relationship based on which caching is done. This approach provides better hit rates than LRU while in some cases performing half the number of replacements. It is felt by allowing more files to remain in the cache that the user can be given more options to use. 2.2 Caching Architecture Fig. 1 given below is showed the architecture how query is moved forward from cache side to server side and server side to cache side [5]. priority has to be settled down i.e. which process is occurring first. The concepts of time sharing are raised in the multiprocessing functions and in next generation like hyper threading it gave better result. Intra file relation is basic to enhance the performance in cache 2.3 Cache Mapping Techniques When we want to know how we can enhance the performance of cache, there are two main factors which effect on performance i.e. hit ratio and searching speed.when processor is found the data in cache it map the cache line with memory address and data is found.it is the hit ratio which reflected more data maximum hit ratio. Searching speed depends upon the region on which searching is occurring this will also impact on performance. There are basic three cache mapping a) Direct Mapping b) N-way Set Associative Mapping c) Fully Associative Mapping 3 Results & Discussions Fig. 1: Client server Cache Architecture In semantic cache inter file relationship [6] data or information is obtained from the distributed file system and it is found on priority bases. Some time total file is checked by a very short procedure. It is observed that almost 50 to 60 % file access occurs only in one precursor. In semantic cache intra file relationship is considered better where files are opened before they are closed. It is the concept of time sharing. Mostly it is observed that time sharing concept is raised where there is some 3.1 Simulator Working In the designed software (simulator) semantic based technique was used which gave the best efficiency in data finding in the cache. These techniques are used to find the memory addresses and cache line in cache which give the desired data or information. For next data finding processor go to the neighbor of previous address. it found the data. This rule is called 90% to 10 % rule. It was observed that behavior of access pattern was totally changed. The performance of semantic cache would be changed [7] by changing Cache spaces, speed, replacement strategies, Greater storage space and using distributed file system are the main advantages of this semantic Cache technique.. In simulator green color showed the hit ratio of access pattern data. It showed that maximum data found in the semantic based cache and generated 70 to 80% data fetching or finding for the required places while red

3 color is showed the miss ratio,(inter\intra file semantics) to perform the functioning or hoarding [1]. This actually did not measure exact ratio which one is better, but here simulator cleared this fact which one is better which is intra file technique [6]. For intra file relationship data or information is obtained from the distributed file systems and its own file traces activity is mostly on precursor. Some time total file is checked by a very short time. It is observed that almost more than 50 % file access is occurring in only one precursor. Given below Fig. 2 showed the working of simulator (software) this compared the performance between discussed Fig. 2: Cache simulator (Software) Semantic Table Address Intra file Inter File Fully Associative Call Hit Data Hit Ratio Hit Data Hit Ratio Hit Data Hit Ratio % 7 47% 3 20% % 11 53% 4 19% % 13 52% 5 20% % 20 67% 7 27% % 20 62% 10 30%

4 3.1.1 Statistical Analysis of Intra File This graph shows that as we increase more addresses, the hit ratio of the data also increases. At a particular point maximum hit ratio is 80% of the performance. here it is compared with semantic cache The average hit ratio of Intra File ( )/5 =74.5% almost equal to 75% Statistical Analysis of Inter File This is histogram diagram for inter file semantic cache technique which shows the 50% to 70 % hit ratio in the semantic cache [8] describes that intra file has less hit ratio as compared to others but in this research it has more hit ratio as compared to fully associative technique. The average hit ratio of fully associative technique ( )/5 =Almost 23%. Comparison of these three techniques it is showed that intra file is best for performance point of view having almost 75% hit ratio while inter file (56% hit ratio) has less as compared to intra file (75% hit ratio) while last technique (fully associative) has least performance as compared to above two which has hit ratio almost 23% as very low other than two. The given Pie graph shows the description of the hit ration between these three techniques which is measured under simulatior results Pie Diagram of Inter File, Intra File and Fully Associative Pie diagram show the comparision between these three The average hit ratio of inter file ( )/5 =56 % Statistical Analysis of Fully Associative In the fully associative technique simulator gives the hit ratio by almost 20% to 30% which is the maximum efficiency results for this technique. The results observed by [9] were different from these results. Because Intra file=parallel lines Area Inter file=net Lines Area Fully Associative=Dotted Area 4 Future Work This research focus on the semantic cache techniques and defined a comparison

5 between inter file, intra file and fully associative and derived a comparative best performance for these calculated results. For future Extension, this area requires the improvement in the semantic cache implementation technique, like grammar oriented searching algorithms, Best search technique, Greedy Algorithm and specific size for cache region. References: [1] Geoffrey H. Kuenning, Gerald J. Popek (1997) Automated hoarding for mobile computers, ACM SIGOPS Operating Systems ACM Symposium on Operating Systems Principles Proceedings of the sixteenth ACM symposium on Operating systems principles Saint Malo, France Pages: ISBN: [2] Edith Cohen, Balachander Krishnamurthy, Jennifer Rexford (1998) improving end to end performance of web using server volumes and proxy filters,acm SIGCOMM Computer Communication Review, Proceedings of the ACM SIGCOMM '98 conference on Applications, technologies, architectures, and protocols for computer communication SIGCOMM '98, Volume 28 Issue 4 Vancouver, British Columbia, Canada Pages: ISSN: Publisher ACM Press New York, NY, USA [3] James Larkby-Lahet, Ganesh Santhanakrishnan, Ahmed Amer, Panos K. Chrysanthis (2005) Self-Tuning Energy-safe Predictors, International Conference On Mobile Data Management Proceedings of the 6th international conference on Mobile data management Ayia Napa, Cyprus SESSION: Caching and replication Pages: ISBN: [4] Maria R. Ebling, Bonnie E. John, M. Satyanarayanan (2002) The importance of trancelucent cache in mobile computing system, ACM Transactions on Computer-Human Interaction (TOCHI) Volume 9, Issue 1 Pages:42 67Publishr ACM NY USA ISSN: [5] David J. Dewitt, Michael J. Franklin, Michael J. Zwilling, C. K. Tan, Michael J. Carey, (1992) Crash Recovery in Client-Server EXODUS ACM SIGMOD International Conference on the Management of Data, Pages: ISBN: San Diego California, USA. [6] Brian D. Noble, M. Satyanarayanan, Dushyanth Narayanan, James Eric Tilton, Jason Flinn, Kevin R. Walker (1997) Agile application-aware adaptation for mobility, ACM Symposium on Operating Systems Principles Proceedings of the sixteenth ACM symposium on Operating systems principles Saint Malo, France Pages: ACM Press New York, NY, USA ISSN: [7] Bjorn por, Jonsson, Mar ia Arinbjarnar, Bjarnsteinn Porsson,Michael j. Franklin,Divesh Srivastava (2006) Performance and Overhead of Semantic Cache Management, ACM Transactions on Internet Technology (TOIT) Volume 6, Issue 3 (August 2006) Pages: ACM Press New York, NY, USA ISSN: [8] Sharun Santhosh,Weisong Shi, Hanping Lufei (2004). An Adaptive Distributed File System for Heterogeneous Network Environments, Proceedings of the Parallel and Distributed Systems, Tenth International Conference on (ICPADS'04) - Volume 00 Page: 145, IEEE Computer Society Washington, DC, USA, ISBN ~ ISSN: , [9] Shaul Dar, Raghu Ramakrishnan (1994) A performance study of transitive closure algorithms, International Conference on Management of Data Proceedings of the A CM SIGMOD international conference on Management of data Pages: Minneapolis, Minnesota, United States