ElasticSearch Communication & Stakeholder Relations 2 — Questions and Answers
Question 1: A product manager asks why Elasticsearch returned different result counts on two identical queries run minutes apart. What is the most accurate explanation to give?
- Elasticsearch uses eventual consistency, so newly indexed documents may not yet be visible in search (Correct answer)
- The queries were not truly identical because HTTP headers differ
- Elasticsearch randomizes result counts to prevent caching
- The cluster automatically rebalances between queries
Correct answer: Elasticsearch uses eventual consistency, so newly indexed documents may not yet be visible in search
Elasticsearch is eventually consistent—documents indexed after the last refresh interval may not appear in search results immediately.
Question 2: A stakeholder requests a guarantee that all documents written to Elasticsearch will never be lost. How should you respond?
- Guarantee full durability by enabling synchronous flushing on every write
- Explain that with proper replica configuration and snapshot policies, data loss risk is minimized but no absolute guarantee exists without external backups (Correct answer)
- Confirm Elasticsearch provides ACID guarantees like a relational database
- Recommend storing all data in Elasticsearch's built-in persistent queue
Correct answer: Explain that with proper replica configuration and snapshot policies, data loss risk is minimized but no absolute guarantee exists without external backups
Elasticsearch reduces data loss risk through replicas and snapshots, but communicating realistic expectations about distributed system guarantees is essential.
Question 3: During a stakeholder review, you need to explain index aliases. Which analogy best communicates their purpose to a non-technical audience?
- An alias is like a username that maps to multiple email inboxes, letting you query several indexes at once without changing application code (Correct answer)
- An alias is a backup copy of an index stored on a different node
- An alias compresses index data to reduce storage costs
- An alias is an Elasticsearch term for a shard
Correct answer: An alias is like a username that maps to multiple email inboxes, letting you query several indexes at once without changing application code
Aliases act as named pointers to one or more indexes, allowing seamless routing and zero-downtime index swaps without application changes.
Question 4: A business stakeholder wants weekly reports on cluster health. Which Elasticsearch metric is most meaningful to include for a non-technical audience?
- JVM heap used percentage
- Cluster health status (green/yellow/red) with a plain-language explanation of what each color means (Correct answer)
- Raw segment merge rate in bytes per second
- Number of pending tasks in the master queue
Correct answer: Cluster health status (green/yellow/red) with a plain-language explanation of what each color means
Cluster health status provides an immediately interpretable signal that non-technical stakeholders can act on without needing Elasticsearch expertise.
Question 5: A developer team asks you to explain the difference between a search query and a filter in Elasticsearch for caching purposes. What is the key point?
- Queries are cached automatically; filters are never cached
- Filters are cached in the filter cache and do not affect relevance scoring, making them faster for yes/no conditions than full-text queries (Correct answer)
- There is no performance difference between queries and filters in modern Elasticsearch
- Filters run on the master node while queries run on data nodes
Correct answer: Filters are cached in the filter cache and do not affect relevance scoring, making them faster for yes/no conditions than full-text queries
Filter context results are cached in the node query cache and skip scoring, making them more efficient for exact-match or range conditions.
Question 6: A compliance officer asks whether Elasticsearch stores data encrypted at rest by default. What is the correct answer?
- Yes, all data is encrypted with AES-256 by default in all versions
- No, encryption at rest is not provided by Elasticsearch itself but can be achieved through OS-level or hardware-level disk encryption (Correct answer)
- Elasticsearch encrypts index files but not translog files
- Encryption at rest requires disabling replication
Correct answer: No, encryption at rest is not provided by Elasticsearch itself but can be achieved through OS-level or hardware-level disk encryption
Elasticsearch does not natively encrypt data at rest; operators must implement encryption at the filesystem or hardware layer (or use Elastic's X-Pack security features in supported tiers).
Question 7: A stakeholder asks how Elasticsearch handles a node failure mid-write. How do you explain shard replica behavior?
- The write is rolled back and the client must retry from scratch
- If the primary shard acknowledges the write and a replica fails, Elasticsearch promotes another replica and the document is not lost assuming at least one healthy replica exists (Correct answer)
- All writes pause until the failed node is restored
- Elasticsearch buffers writes in a separate queue until the node comes back online
Correct answer: If the primary shard acknowledges the write and a replica fails, Elasticsearch promotes another replica and the document is not lost assuming at least one healthy replica exists
Elasticsearch replicates writes to replica shards, and if a node fails, a replica is promoted to primary so acknowledged writes are preserved.
ElasticSearch Communication & Stakeholder Relations 2
A product manager asks why Elasticsearch returned different result counts on two identical queries run minutes apart.
What is the most accurate explanation to give?