Which technologies can reduce write I/O latency in long-distance active-active scenarios? (Multiple choice)

Question

Options

• AGateway-free design
• BFastWrite
• CRDMA
• DOptimistic lock

• A
  69% (29)
• C
  31% (13)

Explanation

In long-distance active-active storage scenarios, gateway-free design (A) reduces latency by eliminating intermediary hops so hosts write directly to storage - every removed gateway cuts round-trip time across a WAN link. FastWrite (B) acknowledges writes to the host as soon as data lands in cache rather than waiting for physical disk commit, shrinking perceived write latency dramatically. Optimistic locking (D) avoids the costly distributed lock acquisition that pessimistic locking requires - instead of making a cross-site round trip to claim a lock before writing, it proceeds optimistically and only checks for conflicts at commit time, which is far cheaper over high-latency links.

Why C (RDMA) is wrong: RDMA (Remote Direct Memory Access) bypasses CPU overhead for memory-to-memory transfers, but it operates over InfiniBand or RoCE - protocols that only work on low-latency LAN/fabric connections. It does not extend across WAN distances, so it offers no benefit in geographically separated active-active deployments.

Memory tip: Picture the three correct answers as "cut the middleman, cache early, assume peace" - no gateway middleman (A), cache-first acknowledgment (B), and assume no conflict until proven otherwise (D). RDMA is the "local-only" impostor that sounds fast but can't cross a continent.

No community discussion yet for this question.