Faculty Publications
Permanent URI for this communityhttps://idr.nitk.ac.in/handle/123456789/18736
Publications by NITK Faculty
Browse
3 results
Search Results
Item Load Balancing of MongoDB with Tag Aware Sharding(De Gruyter, 2016) Shegokar, P.; Thomas, M.V.; Chandrasekaran, K.In the era of Big Data, traditional databases face many problems to process data effectively. To solve problems we have NoSQL databases which can solve the problems and are becoming very popular. MongoDB is a document oriented NoSQL database which can process large data sets with the help of sharding mechanism. MongoDB supports range-based sharding, hash-based sharding and tag aware sharding. Problem with range-based method is that one of the shard will have more load. In case of hash-based method, we can distribute the load among shards but related data aren’t together. Tag aware sharding method is better than the other two methods. But this method need to be enhanced as tagged data can belong to more shards and balancer have to migrate the data to most appropriate shard, hence we are enhancing the tag aware sharding in MongoDB. Improper distribution of data can’t allow us to use all the benefits of sharding. Tag aware sharding is administrator based method in which tags are mentioned by the administrator. According to tags, data is migrated but we need to balance the load. To solve this problem, Weighted Round Robin (WRR) load balancing algorithm is used and it has improved the writing and querying performance. © 2016 Walter de Gruyter GmbH, Berlin/Boston.Item Load balancing of MongoDB with tag aware sharding(Walter de Gruyter GmbH info@degruyter.com, 2015) Shegokar, P.; Thomas, M.V.; Chandrasekaran, K.In the era of Big Data, traditional databases face many problems to process data effectively. To solve problems we have NoSQL databases which can solve the problems and are becoming very popular. MongoDB is a document oriented NoSQL database which can process large data sets with the help of sharding mechanism. MongoDB supports range-based sharding, hash-based sharding and tag aware sharding. Problem with range-based method is that one of the shard will have more load. In case of hash-based method, we can distribute the load among shards but related data aren't together. Tag aware sharding method is better than the other two methods. But this method need to be enhanced as tagged data can belong to more shards and balancer have to migrate the data to most appropriate shard, hence we are enhancing the tag aware sharding in MongoDB. Improper distribution of data can't allow us to use all the benefits of sharding. Tag aware sharding is administrator based method in which tags are mentioned by the administrator. According to tags, data is migrated but we need to balance the load. To solve this problem, Weighted Round Robin (WRR) load balancing algorithm is used and it has improved the writing and querying performance.Item Cancelable iris template generation using modulo operation(Institute of Electrical and Electronics Engineers Inc., 2017) Prasad, M.V.N.K.; Jyothi, A.; Lasya, K.Templates in conventional iris recognition systems are unprotected and highly vulnerable to numerous security and privacy issues. To overcome these issues, biometric template is morphed before storing. The morphed template is called cancelable template and it is unlinkable and non-invertible. The cancelable can be easily replaced and reissued. In this paper, we proposed a novel approach of protecting biometric template based on modulo operation. The proposed method utilizes consistent bit vector which is generated from pre-aligned IrisCodes. These IrisCodes are generated by applying 1-D Log Gabor filter on the iris images. To confirm the efficiency of proposed approach, experiments are conducted on different iris data-sets. By applying proposed cancelable technique, we achieved Equal Error Rate of 0.54% and 0.86% for CASIA-V 1.0 and CASIA-V3-Interval iris datasets respectively. This proposed method also satisfies revocability, unlinkability and irreversibilty criteria. © 2017 IEEE.