All subjects · Information Technology · WIT13 2 hour written exam on data, enabling technologies and systems development

Distributed systems and their issues

Edexcel IAL Information Technology · WIT13 2 hour written exam on data, enabling technologies and systems development

A distributed system spreads data and processing across many networked computers that work together as one system. You need to know why they are needed and the issues of failure, concurrency, replication and performance (spec 13.2). The June 2025 paper set a 6-mark question in which candidates discussed the failure, concurrency and performance of a distributed database.

Practise this in the app: instant marking and a fix for every mistake. Free, no email needed to start.

Practise this lesson

What a distributed system is and why it is needed

A distributed system is a collection of independent computers (nodes), linked by a network, that coordinate to appear to users as a single system. Examples: cloud services, distributed databases, content delivery networks, online banking, multiplayer games, Big Data clusters.

Why needed:

Failure

Nodes, disks and network links fail independently, and a system may suffer partial failure (some parts working, some not). Problems: detecting that a node has failed rather than being slow; a network partition can split nodes into groups that cannot talk to each other; transactions may be left half complete.

Solutions: redundancy and replication, heartbeat monitoring, automatic failover to a standby node, transaction logs and two-phase commit so a transaction completes everywhere or nowhere.

Concurrency

Many users at different nodes may read and update the same data at the same time. Without control this causes lost updates (two clerks book the last seat; one update overwrites the other) and inconsistent reads.

Solutions: record locking (a user locks a record while updating it), timestamps and versioning, and transaction management so updates are applied in a safe order. Locking can cause deadlock, where two transactions each wait for a record the other holds.

Replication and performance

Replication keeps copies of data on several nodes. Benefits: availability if a node fails, faster local reads, load sharing. Issues: every update must reach every copy; copies can be temporarily out of date (consistency versus speed); conflicting updates must be resolved; more storage and network traffic.

Performance issues: network latency and limited bandwidth between nodes; uneven load (some nodes overloaded); the overhead of locking, replication and coordination messages; queries that must join data held at several sites are slow. Solutions: load balancing, caching, placing data close to the users who use it most, partitioning (sharding) data sensibly.

Worked example

Each node of a distributed database is available 95% of the time, independently. Data is replicated on 3 nodes. What is the probability that at least one copy is available?

  1. P(one node down) = 1 - 0.95 = 0.05
  2. P(all 3 down) = 0.053 = 0.000125
  3. P(at least one up) = 1 - 0.000125 = 0.999875

Answer: 99.9875% availability, compared with 95% for a single copy

Key idea: Distributed system = networked nodes acting as one. Needed for scale, performance, reliability and geography. Issues: failure (partial failure, partitions), concurrency (lost updates, locking, deadlock), replication (keeping copies consistent), performance (latency, load, coordination overhead).

Check yourself

  1. Each replica is available 90% of the time, independently. With 2 replicas, what is the probability at least one is available? Give a decimal.
    Show answer

    0.99. 1 - 0.12 = 0.99.

  2. Two booking agents at different nodes both sell the last concert ticket at the same moment. Which issue is this?
    • Replication delay only
    • Hardware failure
    • Network bandwidth
    • Concurrency
    Show answer

    Concurrency. Simultaneous updates to the same data.

  3. How does record locking address the concurrency problem?
    • It encrypts every record
    • A record being updated cannot be changed by anyone else until the update is finished
    • It makes a copy of every record on every node
    • It deletes records that are changed twice
    Show answer

    A record being updated cannot be changed by anyone else until the update is finished. Exclusive access during update.

  4. What is deadlock?
    • A node that has crashed
    • A network cable that is unplugged
    • Two transactions each wait for a resource the other has locked, so neither can continue
    • A record that has been deleted
    Show answer

    Two transactions each wait for a resource the other has locked, so neither can continue. Circular waiting.

  5. What is the main disadvantage of replicating data across many nodes?
    • Every update must be copied to all replicas, and copies may briefly be inconsistent
    • Data becomes less available
    • Reads become slower everywhere
    • It removes the need for backups
    Show answer

    Every update must be copied to all replicas, and copies may briefly be inconsistent. Consistency and update traffic.

  6. What is a network partition?
    • Dividing a hard disk into sections
    • A firewall rule
    • A failure that splits the nodes into groups that cannot communicate with each other
    • A type of data dictionary
    Show answer

    A failure that splits the nodes into groups that cannot communicate with each other. Nodes are running but cut off from each other.

  7. Explain one reason why a query on a distributed database can be slower than on a single local database.
    Show answer

    The data may be held at several different sites, so it has to be fetched across the network, and network latency and limited bandwidth add delay.. Data at many sites, so network delay.

Practise with instant feedback

Sources

Last updated 8 November 2026. Spotted a mistake? Tell us.