🗓️ 07082025 1049
📎

concurrency_control

Concurrency control ensures correctness and consistency when multiple transactions or threads access shared resources at the same time.

🔥 Goals:

• Prevent conflicts (e.g., dirty reads, lost updates)
• Ensure isolation (ACID property)
• Maintain performance and throughput

🧱 Types of Concurrency Problems

Problem	Description	Example
Lost Update	Two writes overwrite each other	T1 and T2 read the same record and both update it
Dirty Read	Read uncommitted changes	T1 writes, T2 reads before T1 commits
Non-repeatable Read	Same read returns different results	T1 reads, T2 modifies and commits, T1 reads again
Phantom Read	Rows appear/disappear in repeated queries	T1: SELECT *, T2 inserts, T1 re-queries and sees more rows

🧠 Summary Table

	optimistic_locking	pessimistic_locking
Assumption	Low contention	High contention
Concurrency	High	Low
Risk	Retry on failure	Blocking, deadlocks
Overhead	Lightweight	Heavyweight
Common Usage	Web apps, distributed sys	Databases, legacy systems

---

References

• ChatGPT