PACELC
CAP leaves out the everyday trade-off between latency and consistency.
- If partitionIf partition: During P, trade off A and C.
- ElseElse: Without P, compare latency and consistency.
- Operating pointOperating point: Make the decision concrete per access path.
Context
Why this is important
Make latency and consistency visible even without partitions.
Make latency and consistency visible even without partitions.
PACELC: inspect the decision
The shop example uses the existing shared database and seven modules. Any added service or deployment model is a hypothetical extension. On the shop’s order flow, the cart reads the last replicated product price from a regional replica. The team permits brief replication lag for fast local responses, but requires a coordinated price check at final checkout. During a partition, checkout waits or fails while the catalogue remains readable. PACELC helps discuss these paths separately and set measurable latency and consistency goals.
This is how you read the picture
- If partition If partition
- Else Else
- Operating point Operating point
| Concept | Meaning | Review question |
|---|---|---|
| If partition | During P, trade off A and C. | Which trade-off applies to price reads during normal operation? |
| Else | Without P, compare latency and consistency. | What changes before checkout? |
| Operating point | Make the decision concrete per access path. | Which additional situation does PACELC describe? |
Benefits & limitations
Helps if:
- Make latency and consistency visible even without partitions.
- Examine read and write paths in normal operation and during partitions.
Boundaries
- PACELC frames trade-offs but proves no property of a specific system.
- PACELC structures design choices but provides no latency or consistency measurements without measurement.
In practice
- Classify partition behavior and normal operation separately for cart display and checkout; identify where availability, latency, and consistency are traded off.
- Set the maximum allowed replication lag for the displayed price and distinguish it from the coordinated price validation at authoritative order submission.
- Measure latency and replication lag per path and test both partition and normal operation; revise the policy if agreed limits for checkout or display are missed.
Read the detailed explanation
PACELC extends CAP reasoning to normal operation: during a partition (P), A represents availability and C consistency; else (E), the trade-off is between latency (L) and consistency (C).
The model highlights the everyday cost of replication and coordination. It summarizes system behavior but does not replace analysis of individual read/write paths, quorums, and failure scenarios.
The shop example uses the existing shared database and seven modules. Any added service or deployment model is a hypothetical extension. On the shop’s order flow, the cart reads the last replicated product price from a regional replica. The team permits brief replication lag for fast local responses, but requires a coordinated price check at final checkout. During a partition, checkout waits or fails while the catalogue remains readable. PACELC helps discuss these paths separately and set measurable latency and consistency goals. cap and cap theorem help compare this decision with adjacent concerns.
Check your understanding
Test yourself in three stages
ReadingCheck in the picture
Which trade-off applies to price reads during normal operation?
View the solution with justification
Regional latency is prioritized for fast price reads.
ApplyingTransferred to a new situation
What changes before checkout?
View the solution with justification
Price coordination happens centrally before the order becomes final.
EvaluateWeighing an alternative
Which additional situation does PACELC describe?
View the solution with justification
The latency-consistency trade-off during normal operation, even without a partition.
Explore further
These building blocks continue the thoughts of this page.
References
- Consistency Tradeoffs in Modern Distributed Database System Design (opens in new tab)Daniel Abadi's primary paper on PACELC and normal operation.
- Perspectives on the CAP Theorem (opens in new tab)Explains CAP guarantees and distributed-system assumptions.
- Brewer’s conjecture and the feasibility of consistent, available, partition-tolerant web services (opens in new tab)Formal primary proof of the partition theorem.