Recoverability and cascadeless schedules in DBMS Theory - Time & Space Complexity

Time Complexity: Recoverability and cascadeless schedules

O(n x m)

Understanding Time Complexity

We want to understand how the time to check or enforce recoverability and cascadeless properties grows as the number of transactions increases.

How does the cost change when more transactions and operations are involved?

Scenario Under Consideration

Analyze the time complexity of checking if a schedule is recoverable and cascadeless.

-- Assume schedule S with n transactions and m operations
FOR each transaction T in S LOOP
  FOR each operation O in T LOOP
    IF O reads data written by another transaction U THEN
      CHECK if U commits before T commits
      IF not, mark schedule as not recoverable
      CHECK if O reads only committed data
      IF not, mark schedule as not cascadeless
    END IF
  END LOOP
END LOOP
    

This code checks dependencies between transactions to ensure recoverability and cascadelessness.

Identify Repeating Operations

Look for loops and repeated checks.

• Primary operation: Nested loops over transactions and their operations.
• How many times: For each of the n transactions, it checks m operations, and for each operation, it may check dependencies with other transactions.

How Execution Grows With Input

As the number of transactions (n) and operations (m) grow, the checks increase roughly by the product of these.

Input Size (n transactions)	Approx. Operations (checks)
10	About 10 x m checks
100	About 100 x m checks
1000	About 1000 x m checks

Pattern observation: The number of checks grows linearly with the number of transactions and operations.

Final Time Complexity

Time Complexity: O(n x m)

This means the time to verify recoverability and cascadelessness grows proportionally with the number of transactions and their operations.

Common Mistake

[X] Wrong: "Checking one transaction is enough to know if the whole schedule is recoverable or cascadeless."

[OK] Correct: Each transaction can depend on others, so all must be checked to ensure no hidden conflicts or uncommitted reads.

Interview Connect

Understanding how the cost grows when checking transaction schedules helps you reason about database reliability and concurrency control in real systems.

Self-Check

"What if we only checked transactions that read from others, skipping those that write only? How would the time complexity change?"

Practice

1. What does a recoverable schedule in a database ensure?

easy

A. Transactions can be undone safely without causing inconsistency

B. All transactions run at the same time without waiting

C. Data is always encrypted during transactions

D. Transactions never read data from other transactions

Solution

1. Step 1: Understand recoverable schedule meaning

   A recoverable schedule ensures that if a transaction needs to be undone, it can be done safely without causing database inconsistency.

2. Step 2: Compare options with definition

   Transactions can be undone safely without causing inconsistency matches this definition exactly, while others describe unrelated concepts.

3. Final Answer:

   Transactions can be undone safely without causing inconsistency -> Option A

4. Quick Check:

   Recoverable schedule = safe undo [OK]

Hint: Recoverable means safe undo of transactions [OK]

Common Mistakes:

• Confusing recoverability with concurrency
• Thinking recoverable means no waiting
• Mixing encryption with recoverability

2. Which of the following is a correct property of a cascadeless schedule?

easy

A. Transactions never write data

B. Transactions can read uncommitted data from others

C. All transactions commit at the same time

D. No transaction reads data written by an uncommitted transaction

Solution

1. Step 1: Define cascadeless schedule property

   A cascadeless schedule prevents transactions from reading data that is not yet committed by other transactions.

2. Step 2: Match options to definition

   No transaction reads data written by an uncommitted transaction states no reading of uncommitted data, which is the key property of cascadeless schedules.

3. Final Answer:

   No transaction reads data written by an uncommitted transaction -> Option D

4. Quick Check:

   Cascadeless = no uncommitted reads [OK]

Hint: Cascadeless means no reading uncommitted data [OK]

Common Mistakes:

• Assuming cascadeless means all commit together
• Confusing cascadeless with recoverability
• Thinking transactions never write data

3. Consider the schedule:
T1: R(A), W(A), commit
T2: R(A), W(A), commit
Is this schedule recoverable and cascadeless?

medium

A. Both recoverable and cascadeless

B. Recoverable but not cascadeless

C. Cascadeless but not recoverable

D. Neither recoverable nor cascadeless

Solution

1. Step 1: Analyze reading and writing order

   T2 reads A before T1 commits. Since T1 has not committed before T2 reads, T2 reads uncommitted data.

2. Step 2: Check recoverability and cascadelessness

   Because T2 reads uncommitted data, the schedule is recoverable (since T1 commits before T2 commits) but not cascadeless.

3. Final Answer:

   Recoverable but not cascadeless -> Option B

4. Quick Check:

   Uncommitted read = not cascadeless, but commit order ensures recoverable [OK]

Hint: Check if reads happen after commit for cascadeless [OK]

Common Mistakes:

• Assuming reading after write means uncommitted read
• Confusing commit order with transaction order
• Ignoring commit before read

4. Identify the problem in this schedule:
T1: W(A), R(B), commit
T2: R(A), W(B), commit
Is this schedule cascadeless?

medium

A. Yes, because no transaction reads uncommitted data

B. No, because T1 reads B after T2 writes

C. No, because T2 reads A before T1 commits

D. Yes, because both commit at the end

Solution

1. Step 1: Check read of A by T2

   T2 reads A which was written by T1, but T1 has not committed before T2 reads A.

2. Step 2: Determine cascadeless property

   Since T2 reads uncommitted data from T1, the schedule is not cascadeless.

3. Final Answer:

   No, because T2 reads A before T1 commits -> Option C

4. Quick Check:

   Uncommitted read = not cascadeless [OK]

Hint: Check if reads happen before commit for cascadeless [OK]

Common Mistakes:

• Assuming commit order doesn't matter
• Thinking commit at end means cascadeless
• Ignoring read of uncommitted data

5. Given transactions:
T1: W(A), commit
T2: R(A), W(B), commit
T3: R(B), W(C), commit
Which schedule below is cascadeless and recoverable, ensuring T2 reads the value written by T1 and T3 reads the value written by T2?

A) T1 -> T2 -> T3
B) T2 -> T1 -> T3
C) T3 -> T2 -> T1
D) T1 -> T3 -> T2

hard

A. T1 -> T2 -> T3

B. T2 -> T1 -> T3

C. T3 -> T2 -> T1

D. T1 -> T3 -> T2

Solution

1. Step 1: Analyze schedule order T1 -> T2 -> T3

   T1 writes A and commits before T2 reads A. T2 writes B and commits before T3 reads B. So no transaction reads uncommitted data.

2. Step 2: Check recoverability and cascadelessness

   Since all reads happen after the respective writes commit, the schedule is both recoverable and cascadeless.

3. Final Answer:

   T1 -> T2 -> T3 -> Option A

4. Quick Check:

   Reads after commit = recoverable & cascadeless [OK]

Hint: Order transactions so reads follow commits [OK]

Common Mistakes:

• Ignoring commit order before reads
• Assuming any order is cascadeless
• Mixing up transaction dependencies