Anil Rawat

Virtual Computing Grid Using Resource Pooling

Cluster and grid computing are becoming very popular, particularly for scientific and engineering applications. Higher bandwidth of computer networks has further boosted development of cluster computing. Local and geographically dispersed deployment of computing resources is the most prominent differentiating issue between a cluster and grid computing based setups. Cluster necessarily requires deployment of multiple nodes at a single location connected via a high speed switch, while grid computing is based on access to remote computing resources, which may be a cluster. Integrating software with multi-layer integration has been the main thrust of research and development in the field of cluster and grid computing. In this paper a novel idea of building a virtual grid over a local area network, deploying resources by pooling has been suggested. The system uses one monitoring server to keep track of various parameters pertaining to the pooled resources and tasks deployed on the system. Software module for integrating the pooled resource has been developed and a prototype model of virtual computing grid using resource pooling (VCGuRP) has been deployed. The paper describes various design and implementation details of the system and also discusses performance issues and their evaluation.

Design, implementation and analysis of multi layer, Multi Factor Authentication (MFA) setup for webmail access in multi trust networks

The obvious advantages of Multi Factor Authentication (MFA) — in terms of enhanced identity theft protection and account hijacking protection — have resulted in its large scale adoption in various security sensitive web applications. In large organizations with different authentication related security policies for Intranet, Extranet and Internet users, varying and multiple authentication implementations are required for achieving security compliance. Layered MFA setups have evolved to facilitate variable authentication implementations in secured web applications. Deploying and handling multi layered, multi factor authentication setups based on Open source freeware solutions is a challenging task for any administrator. This paper presents our work related to design, implementation and integration of a multi layered, multi factor authentication setup for securing the webmail application at our centre, using various freeware open source tools. The implementation ensures fulfillment of varying authentication requirements for Intranet, Internet and Extranet webmail users. Detailed analysis of the implemented setup in terms of compliance to authentication related security policies of our centre have also been presented.

Analysis of process distribution in HPC cluster using HPL

Computing using parallel programming techniques running on High Performance Computing Clusters (HPCC) has become very popular for scientific applications. Evaluation of cluster performance is carried out in many different ways. Memory, interconnect bandwidth, number of cores per processor/ node and job complexity are the major parameters which affect and govern the peak computing power delivered by HPCC. In this paper we carried out experiments using High Performance Linpack (HPL) to analyze effect of job distribution among cores on same processor and among cores on distributed processors. Certain runs of HPL have also been carried out to understand effect of interconnect used for building the cluster. The distributed nature of the HPC cluster using InfiniBand interconnect has been analyzed. Results are represented in form of graphs and corresponding analysis is also included in the paper.

Improved System Components for Secure Data Communication in MANETs using Secured DSR

Efficient and improved methods of secured route discovery and secured data transmission in Mobile Ad-hoc Networks (MANET) are most recent research issues for ensuring full applicability of deploying MANETs in wide range of applications, including strategic as well as non-strategic. Secured route discovery has been addressed by SRP/SMT, as a generic protocol independent technique for secured route discovery. In DSR the issue of secured route discovery is left to be addressed by other protocols. One such concept has been discussed in the proposal of S-DSR, by integrating SRP features in DSR, to achieve secured and multiple route discovery. S-DSR also proposes steps to achieve high throughput, cache management and sustained bandwidth. In this paper impact of proactive route discovery and periodic route validation (as suggested in S-DSR) on SMT/SSP has been analyzed. Authors of SMT/SSP have proposed algorithms for assigning route rating and survival probability. Our proposal in this paper suggests new techniques for assigning route rating and improving path survival probability. These proposed modifications simplify and enhance the functioning of SMT/SSP in terms of better and efficient cache management and improved path survivability.

Enhanced DSR with secured multi-path route discovery and concurrent data transmission

On-demand routing protocols for Mobile Ad-hoc Networks (MANET) include Dynamic Source Routing (DSR), which is capable of discovering multiple routes. Secured Routing Protocol (SRP) on the other hand provides mechanism for secured route discovery, even under adversarial conditions. The paper proposes integration of SRP features in DSR to gain security features for MANET routing requirements for the route discovery phase. Concurrent use of multiple paths for data transmission phase and its implications on performance is also evaluated.

User and Device Tracking in Private Networks by Correlating Logs: A System for Responsive Forensic Analysis

IP address of a device, from where an offending activity was performed, is of limited value, because it does not specify a physical device/user, but an endpoint in network. It is useful to have information about where a device/user was at the time the offending activity was performed. It would be desirable to correlate different pieces of evidence to discover information, such as IP addresses used by the same device, physical address and location of the device, connection time of the device, browsing habits and mail access transactions carried out by the user using the device. Log data from various sources are required to be correlated together to create contexts of information, which is not visible from one source alone. In large networks, users/devices accessing a private network repeatedly can be tracked by analyzing and correlating DHCP, Network Access Control, WWW, Email server logs. With huge amount of logs, the common approach of manual browsing, correlating of log events, based on timelines is tedious, unresponsive approach. Flat file based sequential search system is not responsive, hence RDBMS based tracking systems are desirable. To build a responsive system requires identifying, consolidating log files, conversion, transmission and storage into relational databases. An automated system has been developed at our organization for forensic analysis of network accesses, with device and user tracking as its goal. We present, our approach to perform log management, correlation, which assists in performing responsive forensic analysis of real network with more than 2500 nodes, aimed at tracking users/devices.

Critical analysis of HPL performance under different process distribution patterns

High Performance Linpack (HPL) benchmark is a standard tool to measure the performance of High Performance Computing Clusters (HPCC). There are several parameters which can be optimized to maximize performance delivered by HPCC. HPL provides many such parameters for evaluation of performance. P & Q are two such values which determine the distribution pattern of processes for executing HPL. In this paper HPL runs with different values of P and Q have been carried out for three different values of N while keeping the block size (NB) constant. Value of N represents the problem size of task being executed. Results obtained are included in tables and are also depicted in form of graph and are analytically discussed. These results also provide the basis for deeper analytical approaches for HPCC performance evaluations. This critical analysis also reveals that maximum performance is delivered if the task is evenly distributed over all the available cores and nodes in the HPCC.

Analyitical Study of HPCC Performance using HPL

Performance study of a High Performance Computing Cluster (HPCC) is a complex and multi dimensional activity involving analysis and evaluation of several processes and parameters. Theoretical estimation of peak performance is generally done using simple calculations involving processor clock speed, FLOP rating of processor and number of processors & cores. This estimation typically provides a quantitative number, which is within acceptable limits of variation as against actual numbers. There are many factors which affect the actual performance of a HPCC. In this paper we are presenting results of certain computational experiments carried out on a medium size HPCC. The cluster used for carrying out the experiment is built using six identical server machines interconnected using 1 Gbps Ethernet network switch. To study the performance standard HighPerformance Linpack (HPL) Benchmark has been used with certain variation in number of cores and complexity of task. Intel-optimized version of HPL has been used for carrying out the experiments. Obtained results have been analyzed to understand the behavior of HPC for varying complexity of task. Variation in complexity of task and changing number of processing cores for a particular level of complexity have been studied and presented. Keywords—HPCC, HPL, Peak Performance, Cluster

Performance Analysis of Web based Applications on Single and Multi Core Servers

Performance is one of the most important business requirements for web based applications. The suitability of better platform for any web based application depends on various factors attributable to processor architecture, memory, local/ global network, input/output characteristics etc. Performance testing is an important part of any distributed or web application testing plan. Inclusion of performance estimates into planning and development cycles ensures that the application delivered to the user satisfies sustained performance even at high load, availability and scalability requirements. In this paper we have thrown light on performance comparison of web applications on single and multi core servers. The tests have been carried out on single processor single core server and also on dual processor quad core Xeon based server. The performance comparison was carried out based on response time, throughput (No. of hits v/s response time) and connection pooling v/s no connection pooling. It was found that performance of Java applications running on the multi-core servers is far better than those running on single core servers. Web applications on quad-core server give better performance, high availability and scalability. Apache JMeter was used to load test the functional behaviour and measure performance based on throughput (No of hits v/s response time) for dynamic querying of data using Java servlets on Tomcat web server. Based on the analysis of performance of web server under varying load of no. of users and threads, it was found that applications on quad core server give more throughput and less response time as compared to single core server. Connection pooling increases the performance of web applications by reusing active database connections instead of creating a new connection with every request. Performance was compared by running the tests using Java servlets for connection pooling and also by acquiring connections directly from the JDBC driver without connection pooling. Based on the test results it was found that in case of quad core server, throughput i.e the number of requests served per second is much higher and reuse of connections from pool of connections showed significant improvement in application performance. This paper describes the results of tests carried out for comparing performance (based on response time) between web applications running on single core server and on quad core server. Tests have also been carried out to analyze advantage of connection pooling over no connection pooling. The test results have been shown graphically.

L3C Model of High-Performance Computing Cluster for Scientific Applications

High-performance computing clusters (HPCCs) are widely used for various scientific applications. In a typical scientific research environment, software applications need large but varying number of processing elements and processor cores. To maximize throughput of a computing cluster and optimum utilization of resources, one new model has been proposed. The proposed model visualizes the computing cluster as loosely coupled cluster of clusters (L3C). Execution time for scientific applications also varies in terms of lapsed time for execution and CPU time utilized. The process scheduling algorithm maintains a list of applications to be executed along with respective number of node/core required. Using the L3C model and scheduling algorithm, multiple applications are scheduled on the computing cluster for concurrent execution. Basis for proposing L3C model along with its details is discussed in the paper. Experimental results of performance evaluation of HPC clusters were published earlier by the authors and are referred at respective places. L3C model has certain inherent advantages which are also discussed in the paper.