Prepstellar
Get the app

DP-900 · Common Data Workloads

21 cards

Transactional Workloads

Swipe, scroll or use ← →

  1. Record discrete business events

    A transactional system records transactions that encapsulate specific events an organization wants to track. The unit it deals in is small: a transaction is a small, discrete unit of work.

    Two everyday examples show the scale. Moving money between bank accounts is a transaction. Tracking a retail customer's payment for goods or services is a transaction. Each one is a single, self-contained business event — not a summary, not a history, not a trend.

    That size has consequences. Transactional systems are often high-volume and can handle many millions of transactions in a day, and the data they process must be accessible quickly. Millions of small units, each needing a fast answer, is a very different engineering problem from one large question asked over years of history.

    1 / 21

  2. Quick check

    Which description best defines a transaction in a transactional system?

    1. AA small, discrete unit of work

      Right. A transaction is a small, discrete unit of work that encapsulates a specific event the organization wants to track, such as a bank transfer or a retail payment.

    2. BA preaggregated hierarchy of dimensions and measures

      Preaggregated dimensions belong to an analytical model built for summaries, not to a single recorded business event.

    3. CA vast collection of historical business metrics

      Large volumes of historical metrics describe an analytical store; a transaction is one event, recorded as it happens.

    2 / 21

  3. Name the workload: OLTP

    This kind of work has a name that appears throughout the exam outline: Online Transactional Processing, abbreviated OLTP.

    Keep it apart from the three abbreviations that sit closest to it in the exam outline. Online Analytical Processing is the analytical counterpart, built for querying summaries. Extract, Transform, and Load and Extract, Load, and Transform are movement patterns that carry data toward analysis — they describe how data travels, not how a business event is recorded.

    Term What it names
    OLTP The work transactional systems perform
    OLAP Analytical processing over aggregated data
    ETL and ELT Patterns for moving and preparing data for analysis
    3 / 21

  4. Quick check

    What name is commonly given to the work performed by transactional systems?

    1. AExtract, Transform, and Load

      ETL is a pattern for moving and preparing data toward analysis, not the name of transactional work.

    2. BOnline Transactional Processing

      Right. The work transactional systems perform is called Online Transactional Processing, or OLTP.

    3. COnline Analytical Processing, usually shortened to OLAP

      OLAP names the analytical side of the picture, which queries aggregated data rather than recording events.

    4 / 21

  5. Reads, writes, and the CRUD operations

    An OLTP solution relies on database storage optimized for both read and write operations. Not reads with occasional writes, and not writes that are never read back — both, at speed.

    Transactional records are created, retrieved, updated, and deleted, a set of operations commonly abbreviated CRUD.

    Letter Operation In a retail system
    C Create Record a new payment
    R Retrieve Look up the customer's order
    U Update Correct a delivery address
    D Delete Remove a cancelled line

    These operations are applied transactionally, which is what ensures the integrity of the stored data. The operations describe what happens to a record; applying them transactionally describes the guarantee attached to each change.

    5 / 21

  6. Quick check

    Which expansion identifies the CRUD operations of an OLTP workload?

    1. ACollect, classify, protect, and govern the stored data

      Those are governance activities rather than the record operations an OLTP database performs.

    2. BIngest, cleanse, conform, and summarize the source records

      Ingesting, cleansing, and summarizing describe preparing data for analysis, not the transactional record operations.

    3. CCreate, retrieve, update, and delete database records

      Right. Transactional records are created, retrieved, updated, and deleted, and that set of operations is commonly called CRUD.

    6 / 21

  7. Keep your progress in the app

    That’s 3 of 10 quick checks. In the app they stay answered, and every lesson remembers where you left off.

    Get the app Keep reading here

  8. Live line-of-business applications

    OLTP systems typically support live applications that process business data. These applications are often called line-of-business applications: the point-of-sale terminal, the booking system, the banking front end.

    That word live is the whole workload pattern. The system is not reading through a large historical collection to find a trend; it is recording discrete business events as they happen while keeping the current data available for reads and writes.

    Signal in a requirement Points to
    Record each payment as it is taken Transactional workload
    Millions of small operations a day, answered fast Transactional workload
    Current data must stay readable and writable Transactional workload
    Query years of history for a trend Analytical workload
    7 / 21

  9. Quick check

    Which requirement most clearly indicates a transactional workload?

    1. ASummarizing several years of sales history for a regional trend report

      Summarizing years of history for a trend is analytical work over a large historical collection.

    2. BRecording each retail payment as it happens, with the current data readable and writable

      Right. Recording small, discrete business events while keeping current data available for both reads and writes is the OLTP pattern of a live line-of-business application.

    3. CPreparing a read-mostly copy of business metrics for exploration

      A read-mostly store of metrics prepared for exploration is an analytical store rather than a transactional system.

    8 / 21

  10. ACID: four guarantees on every transaction

    OLTP systems enforce transactions that support ACID semantics. Each letter protects a different aspect of the work, and the fastest way to tell them apart is by the failure each one prevents.

    Property Guarantee The failure it prevents
    Atomicity The transaction succeeds completely or fails completely Half-finished work being left behind
    Consistency The database moves only from one valid state to another An invalid resulting state
    Isolation Concurrent transactions do not interfere with one another A reader mixing before and after
    Durability A committed transaction stays committed Losing work that was already confirmed

    Learn the right-hand column and the letters follow from it.

    9 / 21

  11. ACID: four guarantees on every transaction

    Atomicity treats each transaction as one unit that succeeds completely or fails completely.

    The bank transfer is the standard illustration. Debiting one account and crediting the other must both complete. If either action cannot complete, the other action must fail — the debit is not kept "for now" and retried later, and it is not left standing while the credit is investigated.

    So when a transfer debits Account A but the credit to Account B cannot complete, the correct outcome is that neither change completes. Keeping the debit would leave money that has left one account without arriving anywhere, which is exactly the half-finished state atomicity exists to prevent.

    10 / 21

  12. Quick check

    A bank transfer debits Account A, but the credit to Account B cannot complete. What should an ACID transaction do?

    1. AFail the debit as well, so that neither account change completes

      Right. Atomicity treats the transaction as one unit: if either action cannot complete, the other must fail too.

    2. BPersist the debit and expose both of the account balances

      Leaving the debit in place is the half-finished outcome atomicity prevents, whatever balances are then shown.

    3. CKeep the debit and retry the credit after committing

      A retry after commit would mean the debit had already been kept on its own, which breaks the all-or-nothing rule.

    11 / 21

  13. Consistency keeps the state valid

    Consistency allows a transaction to take the database only from one valid state to another. It is a statement about the states either side of the transaction, not about how the work is split up or who else is reading.

    After a completed bank transfer, the resulting state must reflect the movement of funds from one account to the other: the total is preserved, and both balances tell the same story. A state in which money has been created or destroyed is not a valid state, so a transaction that would produce it cannot be allowed to commit.

    Atomicity asks did the whole unit complete? Consistency asks is the result a state the database is allowed to be in?

    12 / 21

  14. Quick check

    Which ACID property restricts transactions to moving a database from one valid state to another?

    1. AIsolation

      Isolation is about concurrent transactions not interfering with one another, not about the validity of the resulting state.

    2. BThe durability property

      Durability concerns what survives after a commit rather than whether the resulting state is valid.

    3. CConsistency

      Right. Consistency allows a transaction to take the database only from one valid state to another.

    13 / 21

  15. Isolation separates concurrent work

    Isolation prevents concurrent transactions from interfering with one another and requires a consistent database state while they run.

    Return to the transfer. While it is in progress, a concurrent balance check must not combine one account's value from before the transfer with the other account's value from after it. That mixed reading corresponds to no real moment in time: the money would appear twice, or not at all.

    Isolation is the property that rules it out. Note what it is not: nothing has failed halfway here, so it is not atomicity; the transfer's own end state is perfectly valid, so it is not consistency; and nothing committed has been lost, so it is not durability. The fault is purely one transaction seeing another one mid-flight.

    14 / 21

  16. Quick check

    While a bank transfer is in progress, a balance check sees Account A before the transfer and Account B after it. Which ACID property should prevent this?

    1. AThe atomicity property

      Atomicity governs whether the transfer completes as a whole; here the problem is what another reader sees while it runs.

    2. BIsolation

      Right. Isolation prevents concurrent transactions from interfering with one another, so a concurrent balance check cannot mix a value from before the transfer with one from after it.

    3. CDurability

      Durability concerns whether committed work survives afterwards, not what a concurrent reader observes during the transaction.

    15 / 21

  17. Durability preserves what was committed

    Durability means a committed transaction remains committed. Once the transfer completes, the revised balances persist — they are still reflected after the database system is switched off and on again.

    That is the test to apply. A retail payment is committed, and then the system is switched off. When it restarts, the committed payment must still be reflected in the stored data. It is not reversed because no reader was active, it is not left pending until another transaction arrives, and it is not broken apart into separate read and write operations.

    After a commit and a restart Durability says
    The change is still there Correct
    The change was rolled back Violated
    The change is waiting for something else Violated
    16 / 21

  18. Quick check

    A retail payment is committed, and then the database system is switched off. Which outcome demonstrates durability when it restarts?

    1. AThe committed payment is still reflected in the stored data

      Right. Durability means a committed transaction remains committed, so the change persists across the system being switched off and on again.

    2. BThe payment stays pending until another transaction begins

      A committed change does not wait on another transaction; commitment is the point at which it becomes permanent.

    3. CThe payment is reversed because no reader was active

      Nothing about an inactive reader undoes a commit, and reversing it is exactly the loss durability prevents.

    17 / 21

  19. Diagnose the property from the failure

    In practice the question is rarely "define isolation". It is a scenario that has gone wrong, and the task is to name the guarantee that was missing. Use the specific failure condition to pick the property.

    The failure described The property concerned
    Incomplete all-or-nothing work Atomicity
    An invalid resulting state Consistency
    Interference between concurrent work Isolation
    Loss after a commit Durability

    And use the workload pattern the same way. Choose the transactional pattern when the central requirement is to record small, discrete business events with fast access and both reads and writes — the presence of history, aggregation, or trend-finding in a requirement points elsewhere.

    18 / 21

  20. Quick check

    Two orders are processed at the same time, and each one reads a stock level the other is midway through changing. Which guarantee is missing?

    1. ADurability, because a committed change was lost

      Nothing described here was committed and then lost, which is the specific failure durability prevents.

    2. BConsistency, because the resulting state broke a rule the database enforces

      The complaint is not that the final state is invalid; it is what each transaction observed while the other was running.

    3. CIsolation, because concurrent transactions interfered with one another

      Right. Interference between concurrent work is the failure condition that isolation exists to prevent.

    19 / 21

  21. Key takeaways

    • A transaction is a small, discrete unit of work encapsulating an event the organization tracks, and the workload is called OLTP.
    • OLTP storage is optimized for both reads and writes, with records created, retrieved, updated, and deleted — the CRUD operations, applied transactionally to protect data integrity.
    • OLTP supports live line-of-business applications, handling high volumes of small operations that must be answered quickly.
    • ACID names four guarantees: atomicity is all-or-nothing, consistency preserves valid state transitions, isolation separates concurrent transactions, and durability preserves committed changes.
    • Diagnose by the failure: half-finished work, an invalid state, interference between concurrent work, or loss after commit each point to one property.
    20 / 21

  22. Quick check

    Which line pairs each ACID property with the failure it prevents?

    1. AAtomicity prevents interference, isolation prevents loss after a commit, durability prevents half-finished work

      Every pairing is displaced: interference is isolation's concern and half-finished work is atomicity's.

    2. BAtomicity prevents half-finished work, isolation prevents interference, durability prevents loss after commit

      Right. Atomicity makes a transaction all or nothing, isolation keeps concurrent transactions from interfering, and durability keeps a committed transaction committed.

    3. CAtomicity prevents invalid states, isolation prevents half-finished work, durability prevents interference

      Invalid resulting states are the concern of consistency, and half-finished work belongs to atomicity rather than isolation.

    21 / 21

  23. 10 quick checks · then the test

    In the app, finishing the quick checks opens this lesson’s 10-question test, and the ones you miss come back exactly when you’re about to forget them.

    Take this lesson’s test → Next lesson: Analytical WorkloadsGet the app
← Previous Keep your progress · Get the app Next card →

The whole course, on your phone

Lessons you can read, audio you can listen to on the way to work, and practice that remembers what you got wrong.

Get the app