A company is testing Amazon Elasticsearch Service (Amazon ES) as a solution for analyzing system logs from a fleet of Amazon EC2 instances. During the test phase, the domain operates on a…

Question

Options

• AUse a cluster of four data nodes across two AWS Regions. Deploy four dedicated master
• BUse a cluster of six data nodes across three Availability Zones. Use three dedicated master
• CUse a cluster of six data nodes across three Availability Zones. Use six dedicated master
• DUse a cluster of eight data nodes across two Availability Zones. Deploy four master nodes in a

• A
  3% (2)
• B
  83% (53)
• C
  9% (6)
• D
  5% (3)

Explanation

Option B is correct because AWS best practices for highly available Amazon ES require deploying across three Availability Zones (protecting against a full AZ failure) with data nodes in multiples of three (6 nodes = 2 per AZ for even shard distribution), and three dedicated master nodes - an odd number that ensures quorum-based leader election without risk of split-brain.

Why the distractors are wrong:

• A is invalid because Amazon ES is a regional service - you cannot span a single domain across two AWS Regions. Cross-region is not a supported ES cluster topology.
• C uses six dedicated master nodes, which is an even number. AWS explicitly recommends an odd count (3 or 5) for master nodes; an even number breaks quorum mathematics and can cause split-brain scenarios.
• D spans only two AZs, which is insufficient for true high availability (losing one AZ means 50% of capacity is gone), and four master nodes again violates the odd-number rule.

Memory tip: Lock in "3-3-6" - always 3 AZs, always 3 master nodes (odd = quorum), and data nodes in multiples of 3 (e.g., 6) so shards distribute evenly. Any answer with an even master count or cross-region topology is a trap.

No community discussion yet for this question.