Dharavath Ramesh

Implementation of atomicity and snapshot isolation for multi-row transactions on column oriented distributed databases using RDBMS

As traditional relational database have limitation of scalability in the respect of data as well as number of clients. Column oriented databases overcome this problem with the cost of lacking transaction features as compared to relational database. Column oriented databases only assure single row atomic transaction and does not support Snapshot Isolation. In this paper we show how atomicity in multi-row transaction and snapshot isolation can be achieved for column oriented database using RDMBS.

HFIM: a Spark-based hybrid frequent itemset mining algorithm for big data processing

Frequent itemset mining is one of the data mining techniques applied to discover frequent patterns, used in prediction, association rule mining, classification, etc. Apriori algorithm is an iterative algorithm, which is used to find frequent itemsets from transactional dataset. It scans complete dataset in each iteration to generate the large frequent itemsets of different cardinality, which seems better for small data but not feasible for big data. The MapReduce framework provides the distributed environment to run the Apriori on big transactional data. However, MapReduce is not suitable for iterative process and declines the performance. We introduce a novel algorithm named Hybrid Frequent Itemset Mining (HFIM), which utilizes the vertical layout of dataset to solve the problem of scanning the dataset in each iteration. Vertical dataset carries information to find support of each itemsets. Moreover, we also include some enhancements to reduce number of candidate itemsets. The proposed algorithm is implemented over Spark framework, which incorporates the concept of resilient distributed datasets and performs in-memory processing to optimize the execution time of operation. We compare the performance of HFIM with another Spark-based implementation of Apriori algorithm for various datasets. Experimental results show that the HFIM performs better in terms of execution time and space consumption.

Hash Based Incremental Optimistic Concurrency Control Algorithm in Distributed Databases

In this paper, we present a methodology that represents an excellent blossom in the concurrency control environment. It deals with anomalies and assures the reliability of the data before read-write transactions after their successful commitment. This method is based on the calculation of hash value of the data field and compares the current hash value with the previous hash value every time before the write operation takes place. We show that this method overcomes inefficiencies like unnecessary restarts and improves the performance. Finally, this work finds a need for an adaptive optimistic concurrency control method in distributed databases. Thus, a new hash based optimistic concurrency control (HBOCC) approach is presented, where it is estimated to produce reliable results. By performing extensive experiments, we epitomize the performance of this method with existing modalities.

Design of byzantine fault-tolerant transaction commit protocol for heterogeneous distributed databases

In this paper, we present a byzantine transaction commit protocol-based recovery algorithm for distributed database environment. As two-phase commit protocol is restricted due to its blocking nature, it is enhanced to cope with the byzantine coordinator. The proposed protocol can tolerate the fault occurrences to make the transactions complete and takes utmost care towards atomic commit. We also analyse the methodology between different site transactions by analytical performance to make the protocol sufficient. We also perform extensive simulations for choosing better throughput and latency.

Design of a transaction recovery instance based on bi-directional ring election algorithm for crashed coordinator in distributed database systems

In a distributed database environment, when the coordinator site (root node or process) is not working, the environment needs to choose or elect a new one in order to perform the transactional tasks. The elected coordinator takes the lead to perform the activities as well and continues the functioning. If the previous (crashed) site is recovered from the failures then again it leads the system by taking the responsibility. In this paper, a recovery instance based on bi-directional ring election algorithm for the crashed coordinator was brought up. The new algorithm for the recovered site quickly brings the state back by sending messages in parallel instances. This work shows that how the algorithm makes the recovered site faster and takes less time to make the system quickly to handle transactions normally.

Data modelling for discrete time series data using Cassandra and MongoDB

With the current emphasis on big data and its applications, the NoSQL databases have surged in popularity. One of the possible application of the NoSQL databases is the efficient storage of discrete time series data. Cassandra, because of its sequential data storage mechanism and MongoDB, because of its flexible schema and rich query language are ideal fits for storing discrete time series data. In this paper, we present a few data modelling schemes in Cassandra and MongoDB to store the discrete time series data.

A Resilient Failure Evaluation and Patch-up (R-FEP) Algorithm for Heterogeneous Distributed Databases

Blocking methodologies sometimes fail to stop malicious things. Attacks on data oriented applications are a serious threat as per the database management systems concern. The required objective of such environment is to find out the mean time attacks and patch up the failures within the stipulated time. This manuscript represents a failure (attacked) evaluation and patch up instances in distributed database systems. The problems like partition, transaction commitment, and failures state that recovery is much more challenging in databases. This manuscript focuses on the challenges and makes an efficient concern with respect to distributed failure evaluation and recovery.

Preserving atomicity and isolation for multi-row transactions in column-oriented heterogeneous distributed databases

Traditional databases have limitation of scalability with respect to data as well as number of clients. Column-oriented databases have overcome this feature by minimising the cost. Column-oriented databases only ensure single row atomic transaction and does not support snapshot isolation. This paper presents about strong snapshot isolation SI and atomicity for multi-row distributed transactions in HBase. This HBase snapshot isolation uses a novel approach and handles distributed transactions at the end of individual clients. This is also designed to be scalable across large distributed databases in terms of data distribution. Some experiments have been performed extensively to preserve atomicity for distributed transactions in various environments. Experimental results show that the proposed methodology can serve better to preserve atomicity and snapshot isolation in column-oriented HDDBs for multi-row transactions.

An incremental hash-based optimistic concurrency control scenario for failure management in HDDBs - an application approach

During real-time transaction process there is a high probability of transaction failure due to the application of various concurrency control-based protocols and network overload. There are no effective methods to manage these failed transactions that provide the clients an opportunity to renew their transaction in a short span of time. In this paper, we propose a methodology which caters to the need for an effective concurrency control management by providing an efficient way to decrease the real-time as well as non-real-time work done by the server. Our proposed methodology represents an excellent blossom in the failure management environment. This method is based on the calculation of hash value of the data field and compares the current hash value with the previous hash value every time before the write operation takes place. The proposed methodology also allows the client to renew a failed transaction using their previous transaction details. We perform extensive experiments to validate the proposed methodology by using remote method invocation RMI. By performing extensive experiments, we epitomise the performance of this method with existing modalities. Experimental results show that the proposed methodology can achieve atomicity and optimistic instances for failure transactions in an efficient way.

Accelerating Airline Delay Prediction-Based P - CUDA Computing Environment

Machine learning techniques have enabled machines to achieve human-like thinking and learning abilities. The sudden surge in the rate of data production has enabled enormous research opportunities in the field of machine learning to introduce new and improved techniques that deal with the challenging tasks of higher level. However, this rise in size of data quality has introduced a new challenge in this field, regarding the processing of such huge chunks of the dataset in limited available time. To deal such problems, in this paper, we present a parallel method of solving and interpreting the ML problems to achieve the required efficiency in the available time period. To solve this problem, we use CUDA, a GPU-based approach, to modify and accelerate the training and testing phases of machine learning problems. We also emphasize to demonstrate the efficiency achieved via predicting airline delay through both the sequential as well as CUDA-based parallel approach. Experimental results show that the proposed parallel CUDA approach outperforms in terms of its execution time.