Distributed computing

The page you’re looking at is an exam paper for Distributed Computing in the B.E. (AI & DS) curriculum. From the surrounding
content, we can explain the term in the context it’s being tested:

📘 What Distributed Computing Means

Distributed computing refers to a model where computational tasks are spread across multiple machines or nodes that work
together as a system. Instead of relying on a single computer, the workload is divided, coordinated, and executed in parallel, often
over a network.

🔑 Key Themes from the Exam Content

The paper highlights several important aspects of distributed computing:

 Goals of Distributed Systems

o Resource sharing

o Scalability (handling larger workloads by adding more nodes)

o Fault tolerance (system continues working even if some nodes fail)

o Transparency (users shouldn’t feel the complexity of distribution)

 Characteristics of Distributed Systems

o Concurrency (many processes run simultaneously)

o No global clock (nodes operate independently)

o Independent failures (one node can fail without collapsing the whole system)

 Challenges

o Data storage & retrieval: Ensuring consistency and availability across multiple nodes.

o Communication overhead: Extra cost in coordinating tasks across machines.

o Consistency models: Rules for how data updates are seen across the system.

 Applications

o AI & Data Science: Predictive maintenance, healthcare diagnostics, large-scale analytics.

o Cloud platforms: AWS and Microsoft Azure provide distributed computing infrastructure.

o File systems: Hadoop Distributed File System (HDFS) vs. Google File System (GFS).

o Stream processing: Message brokers and real-time data pipelines.

🧩 Why It Matters

Distributed computing is the backbone of modern large-scale systems — from cloud services to big data analytics. It enables handling
massive datasets, running complex AI models, and ensuring reliability in mission-critical applications.

Would you like me to break this down into a study-friendly summary sheet (like quick notes for exam prep), or a conceptual deep
dive with examples from real-world systems like Netflix or Google?

Q1Explain issues related to data storage and retrieval in Al & DS .

Data storage and retrieval in AI and data science (DS) can present several challenges and issues, including:

 Scalability: As the volume of data increases, traditional storage systems may struggle to handle the scale. This can lead to
performance issues and increased costs.

 Data Security: Storing and retrieving sensitive data requires robust security measures to prevent unauthorized access or data
breaches. Compliance with data protection regulations such as GDPR is also crucial.

 Data Quality: Ensuring the quality and integrity of stored data is essential for accurate analysis and modeling. Inaccurate or
incomplete data can lead to flawed insights and decisions.

 Data Retrieval Speed: Quick access to large volumes of data is critical for AI and DS applications. Slow retrieval speeds can
hinder real-time analysis and decision-making.

 Data Integration: AI and DS often require data from multiple sources. Integrating and retrieving data from diverse systems
and formats can be complex and time-consuming.

 Cost Management: Storing and retrieving large volumes of data can be costly, especially with cloud-based solutions.
Optimizing storage and retrieval processes to minimize costs is a significant concern.
 Distributed computing
 Data Governance: Establishing clear policies and procedures for data storage, retrieval, and usage is essential for maintaining
data integrity and compliance.

Q2List & Explain various Characteristics of Distributed Systems.

🔑 Characteristics of Distributed Systems
1. Concurrency
o Multiple processes run simultaneously across different nodes.
o This enables parallel execution and faster task completion, but also introduces synchronization
challenges.
2. No Global Clock
o Each node operates independently without a universal time reference.
o Coordination must rely on logical clocks or synchronization protocols.
3. Independent Failures
o Nodes can fail individually without bringing down the entire system.
o Fault tolerance mechanisms (like replication and recovery) are essential.
4. Resource Sharing
o Distributed systems allow sharing of hardware, software, and data resources.
o This improves efficiency and reduces duplication.
5. Scalability
o Systems can grow by adding more nodes to handle larger workloads.
o Scalability ensures performance remains stable as demand increases.
6. Transparency
o Users should not feel the complexity of distribution.
o Types of transparency include:
 Access transparency (uniform access to resources)
 Location transparency (resources appear local even if remote)
 Replication transparency (users don’t see multiple copies)
 Concurrency transparency (multiple users can access resources without conflict)
 Failure transparency (system continues despite node failures)


Q3 Explain integrating Al and data science in Predictive aintenance and its applications
Here’s a clear explanation of Integrating AI and Data Science in Predictive Maintenance and its applications,
based on the surrounding exam paper content:
🔧 Predictive Maintenance with AI & DS
Predictive Maintenance (PdM) is the practice of using data-driven techniques to anticipate equipment failures
before they occur, allowing organizations to schedule maintenance proactively rather than reactively.
Integration of AI and Data Science:
 Data Collection: Sensors and IoT devices continuously monitor equipment (temperature, vibration,
pressure, etc.).
 Data Storage & Processing: Large volumes of sensor data are stored in distributed systems for analysis.
 Machine Learning Models: AI algorithms detect patterns, anomalies, and trends in the data that indicate
potential failures.
 Distributed computing
 Predictive Analytics: Data science techniques forecast when a machine is likely to fail, enabling timely
intervention.
 Decision Support: Insights are provided to maintenance teams for scheduling repairs, reducing downtime
and costs.

📘 Applications
1. Manufacturing Industry
o Detects wear and tear in machinery.
o Reduces unplanned downtime and improves production efficiency.
2. Automotive & Transportation
o Monitors vehicle components (engines, brakes, tires).
o Predicts failures to enhance safety and reduce breakdowns.
3. Energy Sector
o Tracks turbines, transformers, and power grids.
o Prevents outages by forecasting equipment failures.
4. Healthcare Equipment
o Ensures reliability of critical machines like MRI scanners or ventilators.
o Minimizes risks by predicting malfunctions.
5. Aerospace & Aviation
o Monitors aircraft engines and systems.
o Improves safety and reduces maintenance costs.

Q4 Goals of Distributed System

1. Boosting Performance
The distributed system tries to make things faster by dividing a bigger task into small chunks and finally processing
them simultaneously in different computers. It's just like a group of people working together on a project. For
example, when we try to search for anything on the internet the search engine distributes the work among
several servers and then retrieve the result and display the webpage in a few seconds.
2. Enhancing Reliability
Distributed system ensures reliability by minimizing the load of individual computer failure. If one computer gets
some failure then other computers try to keep the system running smoothly. For Example, when we search for
something in social media if one server gets an issue then also we are able to access photos, and posts because
they switch the server quickly.
3. Scaling for the Future
Distributed systems are experts at handling increased demands. They manage the demands by incorporating
more and more computers into the system. This way they run everything smoothly and can handle more users.
4. Resourceful Utilization
Resource Utilization is one of the most prominent features of a Distributed system. Instead of putting a load on
one computer, they distribute the task among the other available resource. This ensures that work will be done by
utilizing every resource.
Q5 Discuss the use of Al & DS in Healthcare and Medical Diagnostics
Use of AI & Data Science in Healthcare and Medical Diagnostics
 Distributed computing
1. Introduction
Artificial Intelligence (AI) and Data Science (DS) are transforming the healthcare industry by improving diagnosis,
treatment planning, patient monitoring, and hospital management. These technologies analyze large amounts of
medical data to help doctors make faster and more accurate decisions.

2. AI and Data Science in Medical Diagnostics

1. Disease Detection and Diagnosis
AI algorithms can analyze medical data such as X-rays, CT scans, MRI scans, and pathology reports to detect
diseases early.
Examples:
 Detecting cancer tumors in medical images
 Identifying lung diseases from chest X-rays
 Diagnosing skin cancer using image recognition
Benefits:
 Faster diagnosis
 Higher accuracy
 Early disease detection

2. Medical Image Analysis

AI models like deep learning and computer vision are used to analyze medical images.
Applications include:
 Tumor detection
 Brain abnormality detection
 Heart disease diagnosis
This helps radiologists analyze images quickly and reduce human errors.

3. Predictive Analytics for Disease Risk

Data science uses historical patient data to predict possible diseases before they occur.
Examples:
 Predicting diabetes risk
 Predicting heart attacks
 Predicting hospital readmission rates
Hospitals use these predictions to take preventive actions.

4. Personalized Medicine
AI and DS help create personalized treatment plans based on a patient’s:
 Genetic information
 Medical history
 Lifestyle
Example:
Cancer treatments can be customized depending on how a patient’s body responds to certain drugs.
 Distributed computing

5. Clinical Decision Support Systems (CDSS)

AI-powered systems assist doctors in making clinical decisions by analyzing patient records and medical research.
Functions:
 Suggest possible diagnoses
 Recommend treatments
 Alert doctors about drug interactions
This improves patient safety and treatment quality.

6. Remote Patient Monitoring

AI combined with IoT medical devices can monitor patients remotely.
Examples:
 Wearable devices tracking heart rate and blood pressure
 Monitoring diabetes glucose levels
 Tracking sleep and activity levels
Doctors can monitor patients without hospital visits.

7. Drug Discovery and Development

AI helps pharmaceutical companies analyze huge biological datasets to discover new medicines faster.
Benefits:
 Reduces drug development time
 Lowers research costs
 Improves success rates of clinical trials

8. Chatbots and Virtual Health Assistants

AI chatbots provide basic medical assistance and health guidance.
Examples:
 Symptom checking
 Appointment scheduling
 Medication reminders
This reduces workload for healthcare professionals.

3. Advantages of AI & Data Science in Healthcare

 Faster and accurate diagnosis
 Early disease detection
 Reduced medical errors
 Improved patient care
 Cost reduction in healthcare systems
 Efficient hospital management
 Distributed computing
4. Challenges and Limitations
 Data privacy and security issues
 High implementation cost
 Need for high-quality medical data
 Ethical concerns in AI decisions
 Dependence on technology

5. Conclusion
AI and Data Science are revolutionizing healthcare and medical diagnostics. They help doctors diagnose diseases
earlier, provide personalized treatment, and improve overall patient care. Despite some challenges, these
technologies will continue to play a major role in the future of healthcare.

Q6 short notes
1. Distributing Computational Tasks
2. Communication Overhead in Distributed Computing
1. Distributing Computational Tasks
Distributing computational tasks means dividing a large problem into smaller tasks and assigning them to
multiple computers or processors in a distributed system. Each system works on its assigned task simultaneously,
which improves processing speed and efficiency.
Key Points
 Large problems are split into smaller sub-tasks.
 Tasks are executed on different machines or nodes in a network.
 Results from all nodes are combined to produce the final output.
 Helps in parallel processing and reduces execution time.
Advantages
 Faster computation
 Better resource utilization
 Ability to handle large datasets
 Improved system performance
Example
In big data analysis, data is distributed across many servers, and each server processes a portion of the data
simultaneously.

2. Communication Overhead in Distributed Computing

Communication overhead refers to the extra time and resources required for data exchange between different
nodes in a distributed system while performing tasks.
Key Points
 Nodes must communicate with each other to share data and results.
 This communication can cause delays and additional network traffic.
 It includes data transfer time, synchronization time, and message passing.
Causes
 Distributed computing
 Network latency
 Large data transfers
 Frequent communication between nodes
 Poor network bandwidth
Effects
 Increased execution time
 Reduced system efficiency
 Higher network load
Example
When multiple servers process parts of a task, they must send intermediate results to each other, which may
slow down the overall computation.

UNIT 2

Q1 Differentiate between Hadoop Distributed File System (HDFS) and Google File System (GFS).
Difference Between Hadoop Distributed File System (HDFS) and Google File System (GFS)

Aspect HDFS (Hadoop Distributed File System) GFS (Google File System)

Origin Developed by Apache Hadoop (open-source) Developed by Google for internal use

Uses NameNode (master) and DataNodes

Architecture Uses Master and Chunkservers
(workers)

Block / Chunk
Default 128 MB or larger Default 64 MB chunks
Size

Fault Tolerance Data replicated across DataNodes (usually 3 copies) Data replicated across Chunkservers

Big data storage and processing using Hadoop Designed for Google's large-scale
Primary Use
ecosystem applications

Write Pattern Write-once, read-many model Supports frequent appends to files

Proprietary system used internally by

Availability Open-source and widely used in industry
Google

Short Conclusion
HDFS and GFS are both distributed file systems designed to store and process large datasets across multiple
machines. However, HDFS is open-source and optimized for big data processing, while GFS is Google's
proprietary system designed for large-scale internal applications with efficient file appending.

Q2 Explain Cluster Computing

Cluster Computing
Cluster Computing is a type of computing in which multiple computers (nodes) are connected together through
a network and work as a single system to perform tasks. These computers share resources and divide the
workload to improve performance, reliability, and scalability.
Key Features
 Multiple nodes work together to complete tasks.
 Distributed computing
 Nodes are connected through a high-speed network.
 The workload is distributed among different machines.
 Provides high performance and fault tolerance.
Components of Cluster Computing
1. Nodes – Individual computers in the cluster.
2. Network – Connects all nodes for communication.
3. Cluster Management Software – Controls task distribution and monitoring.
4. Storage System – Shared storage for data access.
Advantages
 High processing power
 Improved system reliability
 Scalability (easy to add more nodes)
 Cost-effective compared to supercomputers
Applications
 Scientific research
 Weather forecasting
 Big data processing
 Artificial Intelligence and Machine Learning
Example
A university research lab may use a cluster of computers to process large scientific simulations faster than a single
computer.

Q3 Write note on Eager Replication & Lazy Replication

Eager Replication Lazy Replication
Eager replication and lazy replication are two strategies used in database management systems to ensure data con
sistency across multiple replicas. Here's a brief overview of each:
 Eager Replication: In this approach, updates are propagated and applied to all replicas within the transactio
n boundary. This means that all replicas must be synchronized before any updates are made. This method e
nsures that all replicas are identical and up-to-date at the same time. However, it can be less efficient for lar
ge databases or systems with many replicas, as it requires more resources and can lead to increased transac
tion times.

 Lazy Replication: In lazy replication, updates are propagated from one replica to others asynchronously. Thi
s means that replicas can diverge from each other, allowing for more efficient use of resources and potentia
lly faster updates. However, this also means that replicas may not always be identical, which can lead to inc
onsistencies if not properly managed. Lazy replication is often used in systems where eventual consistency i
s more important than immediate consistency.

⚡ Eager Replication

 Definition: Updates are propagated to all replicas immediately, before the transaction is considered
complete.
 Consistency: Ensures strong consistency—all replicas always reflect the latest state.
 Advantages:
 Distributed computing
o No stale data; users always see up-to-date information.
o Simplifies reasoning about correctness.
 Disadvantages:
o High communication overhead, since every update must be synchronized across replicas instantly.
o Can reduce performance and scalability, especially in geographically distributed systems.
🕒 Lazy Replication
 Definition: Updates are applied to one replica first, and then propagated to others asynchronously (later in
time).
 Consistency: Provides eventual consistency—replicas may temporarily differ, but converge over time.
 Advantages:
o Lower communication overhead.
o Better performance and scalability, especially for read-heavy workloads.
 Disadvantages:
o Risk of stale or inconsistent data until synchronization completes.
o More complex conflict resolution if multiple replicas are updated independently.
🔑 Key Difference

Aspect Eager Replication Lazy Replication

Update Timing Immediate, synchronous Deferred, asynchronous

Consistency Strong consistency Eventual consistency

Performance Higher overhead, slower scalability Faster, more scalable

Use Cases Banking, financial transactions Social media feeds, caching systems

Q4 Explain Consistency Model & its Types

In distributed computing, a consistency model defines the rules and guarantees about how data is read and
written across multiple nodes in a system. Since distributed systems often replicate data across different
machines, consistency models specify what a client can expect when accessing that data. The exam paper you’re
viewing highlights this as a key concept in understanding how distributed systems behave.
🔑 Consistency Model Explained
 It describes the agreement between processes on the values of shared data.
 Ensures that despite replication and concurrent updates, the system provides predictable results to users.
 Different models balance performance, availability, and correctness depending on the system’s design.
📚 Types of Consistency Models
Here are the major types typically discussed:
1. Strict Consistency
o Every read returns the most recent write.
o Impractical in real distributed systems due to network delays.
2. Sequential Consistency
o All processes see operations in the same order, though not necessarily in real-time.
o Easier to implement than strict consistency.
3. Causal Consistency
 Distributed computing
o If one operation causally affects another, all processes must see them in that order.
o Independent operations may be seen in different orders.
4. Eventual Consistency
o Common in large-scale systems (e.g., DNS, NoSQL databases).
o Updates will eventually propagate so all replicas converge, but temporary inconsistencies are
allowed.
5. Weak Consistency
o No guarantee that reads reflect the latest writes.
o Often used when performance is prioritized over accuracy.
6. Release Consistency / Entry Consistency
o Specialized models used in shared-memory systems.
o Synchronization points determine when consistency is enforced.

Q 5 Differentiate between AWS & Microsoft Azure

Based on the exam paper you’re viewing, the question asks you to differentiate between AWS (Amazon Web
Services) and Microsoft Azure in the context of distributed computing platforms. Here’s a structured explanation
aligned with the surrounding content:

🌐 AWS vs. Microsoft Azure

Aspect AWS Microsoft Azure

Provider Amazon Microsoft

Launch Year 2006 2010

Broadest range of services, global infrastructure, Deep integration with Microsoft products,
Core Strengths
strong in IaaS (Infrastructure as a Service) strong in PaaS (Platform as a Service)

Compute Virtual Machines, Azure Functions, and strong

EC2 (Elastic Compute Cloud) for virtual servers
Services hybrid cloud support

Blob Storage, Disk Storage, and integration

Storage S3 (Simple Storage Service), Glacier for archival
with SQL Server

Networking VPC (Virtual Private Cloud), Direct Connect Virtual Network, ExpressRoute

AI & Data Azure Machine Learning, Cognitive Services,

SageMaker, Rekognition, Redshift
Services Synapse Analytics

Pay-as-you-go, reserved instances, hybrid

Pricing Model Pay-as-you-go, reserved instances, spot pricing
benefits for Windows licenses

Market Second largest, strong enterprise adoption

Largest cloud provider globally
Position due to Microsoft ecosystem

🔑 Key Differentiation Points

 Ecosystem Integration:
o AWS is more agnostic, appealing to startups and enterprises across diverse tech stacks.
o Azure is tightly integrated with Windows Server, Active Directory, and Office 365, making it
attractive for organizations already invested in Microsoft tools.
 Hybrid Cloud:
 Distributed computing
o AWS focuses on public cloud dominance.
o Azure emphasizes hybrid solutions (Azure Arc, Azure Stack) to bridge on-premises and cloud
environments.
 Global Reach:
o AWS has the widest global infrastructure with more regions and availability zones.
o Azure is rapidly expanding but often leverages Microsoft’s enterprise relationships.

1. Computer Services

Elastic
AWS EC2 Beanstalk AWS Lambda (Serverless)
(PAAS)

Virtual App Service Azure

Azure
Machine (PAAS) Functions (Serverless)

2. Networking

Direct Route Cloud

AWS VPC ELB
Connect 53 Front

Virtual Azure Azure Load

Azure ExpressRoute CDN
Network DNS Balancer

3. Storage Services

Amazon
AWS S3 SQS EFS
Glacier

Blob Queue File Azure

Azure
Storage storage Storage Backup

4. Database Services

AWS RDS Dynamo DB Redshift

Document SQL Data

Azure SQL
DB Warehouse

Q6 Explain Message Brokers and Stream Processing in Distributed Computing.

In distributed computing, Message Brokers and Stream Processing are two important mechanisms for handling
communication and data flow across systems. The exam paper you’re viewing highlights them as part of the
broader discussion on cloud platforms and distributed architectures.

📬 Message Brokers
 Definition: Middleware systems that manage communication between distributed applications by
receiving, routing, and delivering messages.
 Role in Distributed Computing:
o Decouple producers (senders) and consumers (receivers).
o Ensure reliable delivery even if one component is temporarily unavailable.
o Support asynchronous communication, improving scalability.
 Examples: Apache Kafka, RabbitMQ, ActiveMQ.
 Key Features:
 Distributed computing
o Message queuing (FIFO delivery).
o Publish/subscribe model (multiple consumers can receive the same message).
o Persistence and fault tolerance.

🔄 Stream Processing
 Definition: Real-time processing of continuous flows of data (streams) as they arrive.
 Role in Distributed Computing:
o Enables immediate insights and actions on incoming data.
o Handles high-throughput, low-latency workloads.
o Often used in analytics, monitoring, and event-driven applications.
 Examples: Apache Flink, Apache Spark Streaming, Kafka Streams.
 Key Features:
o Continuous computation rather than batch.
o Windowing (grouping data over time intervals).
o Fault tolerance and scalability across distributed nodes.

⚖️Difference in Focus
 Message Brokers → Focus on transport and delivery of messages reliably.
 Stream Processing → Focus on real-time computation and transformation of data streams.