Chapter # 11

 

Fault Tolerance

Fault tolerance is the system’s ability to continue functioning without interruption, even when one or more of its components fail. It ensures that service remains available, recovers from errors completely and transparently, and handles malfunctions in a way that users may not even notice.
However, fault tolerance comes with trade-offs:

  • It may slow down the system,

  • Require more storage and hardware, and

  • Increase operational costs.

Thus, fault tolerance is always a balance between cost and the desired level of reliability.

Failure vs Error

These terms are related but not the same:

  • Error: A defect or flaw in the system's logic or state that might lead to a failure. Errors can exist without causing failure if they are caught or corrected in time.

  • Failure: Happens when the system deviates from its expected behavior, such as becoming unreachable or giving wrong results. A failure is the visible result of one or more errors.

In summary:
Error → may lead to → Failure

Phases of Fault Tolerance

Fault-tolerant systems usually go through the following phases:

  1. Error Detection
    The system must first identify that something is wrong.

  2. Damage limitation
    Prevent the error from spreading to other parts of the system.

  3. Error Recovery
    Fix the error or restore the system to a stable state to avoid failure.



Processor Faults

Processor faults occur when a processor behaves in an unexpected or incorrect manner. These faults can be categorized into three types:

  1. Fail-Stop

    • The processor completely stops functioning and never responds again.

    • Other processors can detect that it has failed.

  2. Slowdown

    • The processor continues to run, but in a degraded (slower) state.

    • It might eventually fail completely.

  3. Byzantine Fault

    • The processor may:

      • Fail intermittently,

      • Run slowly or incorrectly, or

      • Appear to function normally while giving incorrect outputs on purpose.

    • This is the most unpredictable and difficult type to detect.

Network Faults

Network faults occur when processors are unable to communicate properly with one another. Common types include:

  1. One-Way Link

    • A communication link where one processor can send messages, but the other cannot receive them.

  2. Network Partition

    • A situation where a portion of the network becomes isolated, meaning that some processors can no longer communicate with others in a different segment.

Attributes of a Fault-Tolerant System

A fault-tolerant system is designed to operate reliably even when parts of it fail. Key attributes include:

  1. Availability

    • The system is ready and able to perform its functions at a given moment.

    • A highly available system is accessible and operational almost all the time.

  2. Reliability

    • The ability of the system to operate continuously without failure.

    • Measured over a period of time, not just at a single moment.

  3. Safety

    • Even if the system fails, it does so in a way that doesn’t cause major harm or dangerous outcomes.

    • It also prevents the fault from affecting other parts of the system.

  4. Maintainability

    • Faults or failures in the system can be easily detected, diagnosed, and fixed.












Types of Failure in Distributed Systems

Type of FailureDescription
Crash FailureThe server completely halts, but all operations before the crash were performed correctly.
Omission FailureThe server fails to respond to requests. It can be broken down into two subtypes:
→ Receive OmissionThe server fails to receive incoming messages.
→ Send OmissionThe server fails to send outgoing messages.
Timing FailureThe server responds, but outside the expected time interval (either too late or too early).
Response FailureThe server responds, but the response is incorrect. This includes:
→ Value FailureThe actual value returned is wrong.
→ State Transition FailureThe server does not follow the correct control flow or state sequence.
Arbitrary Failure (Byzantine)The server may give completely unpredictable or incorrect outputs, at any time. This is the most severe and difficult to handle failure type.






Fault Tolerance Mechanisms in Distributed Systems

Distributed systems use various techniques to tolerate and recover from faults:

1. Replication-Based Fault Tolerance Technique

  • Data or services are replicated across multiple machines or servers.

  • Ensures that if one server fails, another can take over without stopping the entire system.


2. Process-Level Redundancy (PLR) Techniques


PLR means running extra copies of a process to catch and fix problems.

Main Points:

  • Works best for temporary (transient) faults that fix themselves.

  • Uses software tools to handle errors.

  • The system takes regular checkpoints (saves its state), so it can go back if something goes wrong.

  • PLR compares results from different process copies to make sure everything is running correctly.



3. Fusion-Based Redundancy Technique

  • This method is a cheaper and smarter alternative to full backups.

    Main Points:

    • Uses fewer backup machines by combining them to cover many systems.

    • Costs less than making full copies of everything.

    • Good for systems where failures don't happen often.

    • Recovery takes more time and effort, but it’s worth it when problems are rare.



Comments

Post a Comment

Popular posts from this blog

PDC syllabus

Chapter 7: Scheduling in multiprocessing multi processing scheduling Categories of multiprocessor Technique of multiprocessor sheduling symmnetric and asymmetric Symmetric scheduling main global queues and private queues diff Asymmetrc multiprocessor process infininty  soft inifnity hard infinity load balancing static and dynanmic and distribued or in 3no ki aaga 2 2 types hain Issues in load balancing Chapter 3: Processess and thread difference  user level and kernel level thread Single thread process and multi thread process kaisa work krta Multi tasking or multi threading Multi threading programming Scaller processor / super scaller processor Intel hyper threading 1st slide: Taxonomy Pf propsed this ( MCQs ka liya yeh slide ) 4 taxonoy - Sist, simd, mist wali yeh short ka lia diff bhi aasakta ha kissi aik ka bara ma bhi Chapter#1: Short: Parallel and serial Computing diff Real life examples ka sath HPC and HTC diff Types of Parallel computing system and their diff Distribut...
Read more

PDC

Parallel and series computing real life example: Serial computing : Imagine your computer is downloading a file . First, it downloads the file. After it’s fully downloaded, it scans it for viruses. One task after another — step by step. Parallel computing : Now imagine your computer is very smart . While it is still downloading the file, It starts scanning the parts that are already downloaded. Both tasks happen at the same time — faster! In short : Serial = download ➔ then scan. Parallel = download and scan together. Distributed system: Distributed Systems A distributed system is a group of independent computers that work together and look like a single system to users. These computers communicate by sending messages to each other. Each computer has its own instruction manager or control unit. Distributed systems are used whenever computers are connected in a network. Uses of Distributed Systems Distributed systems are used in many areas, lik...
Read more

Video 1 The Power of Natural Stones: Why They Matter

This content is based on personal research and traditional crystal knowledge. It is not intended as medical advice, diagnosis, or treatment. Always consult qualified healthcare professionals for any physical or mental health concerns. Welcome to Kreistal Store, where nature and peace meet. Today we're diving into the fascinating world of natural stones and crystals. We'll explore their energy, their purpose, and how they can truly transform the spaces you live in and the way you feel every day. What Makes Natural Stones Special Every stone you see, every crystal you hold, carries something truly unique inside it. These aren't just pretty rocks. They're pieces of the earth that have been forming for millions and millions of years. Deep underground, under intense pressure and heat, minerals come together in specific patterns. This process creates the unique structure and energy signature of each stone. When you hold a piece of amethyst, you're holding something tha...
Read more