JNTU MCA 2020 BIG DATA ANALYTICS Previous Question Paper (October/ November - 2020 - R17)

Parallel computing is a type of computation where many calculations or the execution of processes are carried out simultaneously. It relies on the principle that large problems can often be divided into smaller ones, which can then be solved concurrently. In a parallel system, multiple processors work together on a single task, typically sharing a common memory space or communicating via a high-speed interconnect.

A conceptual diagram of parallel computing involves a single control unit managing multiple processors that access a shared memory or distributed memory to process data segments in parallel.

Distributed computing is a model in which components of a software system are located on different networked computers, which communicate and coordinate their actions by passing messages. The system appears to the user as a single coherent system. Unlike parallel computing, distributed computing often deals with geographically dispersed nodes, each having its own private memory and processing power.

A conceptual diagram of distributed computing shows multiple independent nodes (each with its own CPU and memory) connected via a network, where a master node or coordinator manages the distribution of tasks and aggregation of results across the cluster.

Key differences include:

• Architecture: Parallel computing usually occurs within a single machine or tightly coupled cluster; distributed computing spans multiple autonomous machines across a network.
• Memory: Parallel systems often use shared memory; distributed systems use message passing because memory is physically separated.
• Goal: Parallel computing focuses on speed and performance by reducing execution time; distributed computing focuses on scalability, fault tolerance, and resource sharing.

Capitalizing on Big Data is a complex undertaking that involves significant technical, organizational, and strategic hurdles. The primary challenges include:

1. Data Volume and Variety: The sheer scale of data generated at high velocity makes traditional storage and processing systems inadequate. Furthermore, managing unstructured data such as social media feeds, sensor logs, and multimedia requires specialized tools like Hadoop or NoSQL databases, which demand significant infrastructure investment.
2. Data Quality and Integrity: Big Data is often noisy, incomplete, or inconsistent. Ensuring data veracity—the accuracy and reliability of the data—is a major challenge, as poor-quality data leads to flawed analytical insights and incorrect business decisions.
3. Skill Gap: There is a persistent shortage of professionals skilled in data science, machine learning, and Big Data engineering. Organizations struggle to recruit or train staff capable of managing complex data pipelines and interpreting analytical outputs.
4. Data Security and Privacy: Handling massive datasets increases the surface area for cyberattacks. Organizations must navigate stringent regulatory requirements like GDPR, ensuring that sensitive information is protected while still being accessible for analysis.
5. Integration and Silos: Many businesses operate with legacy systems that do not communicate effectively with modern Big Data platforms. Breaking down these organizational and technical silos to create a unified view of data is a time-consuming and costly process.
6. Cost and ROI Uncertainty: Implementing Big Data solutions involves high upfront costs for cloud infrastructure, software licensing, and talent. Demonstrating a clear return on investment (ROI) remains difficult, as the link between data insights and tangible business outcomes is not always immediate.
7. Scalability and Performance: As data grows, systems must scale horizontally without compromising performance. Maintaining low-latency access to real-time data while managing massive batch processing jobs requires sophisticated architectural planning.