Replication and Consistency

Leaders, followers and quorums

With leader-follower replication, writes go to the leader and are copied to followers that serve reads; asynchronous replication can make followers lag, so a user may not see their own write (use read-your-writes routing for that user). Leaderless systems (Dynamo-style) write to W of N replicas and read from R; if R + W > N, read and write sets overlap, so a read sees the latest acknowledged write. The CAP theorem says that during a network partition you must choose between consistency and availability.

Which quorum settings overlap?, run

I ran this with Python 3.12.3 using only the standard library; inputs are fixed or seeded, so the output is reproducible. For N = 3, every combination with R + W greater than 3 guarantees overlap; W = 2 and R = 2 is the common balanced choice.

# Leaderless replication: when do reads see the latest write?
N = 3
for W in range(1, N + 1):
    for R in range(1, N + 1):
        overlap = R + W > N
        note = "strong (overlap)" if overlap else "may read stale"
        print(f"N={N} W={W} R={R}: {note}")

Output:

N=3 W=1 R=1: may read stale
N=3 W=1 R=2: may read stale
N=3 W=1 R=3: strong (overlap)
N=3 W=2 R=1: may read stale
N=3 W=2 R=2: strong (overlap)
N=3 W=2 R=3: strong (overlap)
N=3 W=3 R=1: strong (overlap)
N=3 W=3 R=2: strong (overlap)
N=3 W=3 R=3: strong (overlap)

Name the consistency you need per feature

A like counter can be eventually consistent; an account balance cannot.

Quick check: With N=3 replicas, which setting guarantees reads overlap the latest write?

• W=1, R=1
• W=2, R=2
• W=1, R=2
• W=2, R=1