USE CASE · ELASTICSEARCH REPLACEMENT

ELASTICSEARCH
REPLACEMENT.

You know the operational burden: endless config knobs, JVM tuning, shard calculators, master election storms, and a team of specialists to keep it running. XERJ is a drop-in replacement — same API on port 9200, same query DSL, same client libraries — on one binary instead of a JVM cluster.

REPRODUCIBLE · ES 8.13 · SEARCH & FILTER · SCORECARD.md

XERJ ES 8.13

Bool query 1.38 ms 15.90 ms 11.5×

Query-string query 1.35 ms 9.35 ms 6.9×

Wildcard query 1.35 ms 9.10 ms 6.8×

Range filter 1.34 ms 4.66 ms 3.5×

Match query 0.57 ms 1.45 ms 2.5×

Term query 1.36 ms 2.15 ms 1.6×

P50 · 1M-DOC LLM-TELEMETRY CORPUS · FULL 91-CELL MATRIX AT xerj.org/benchmarks →

THE COST OF ELASTICSEARCH

Memory is the bottleneck. A production cluster spends most of its budget on RAM for the JVM heap, not for your data. Scale to more indices and you're buying heap, not capacity.
Specialized engineers required. ES needs dedicated operators who understand shard allocation, heap sizing, GC tuning, ILM rollover, and cross-cluster replication.
GC pauses cause p95 spikes. JVM stop-the-world pauses inject latency spikes that show up in your SLOs and your on-call pages.
Feature-gated pricing. SSO, RBAC, ML anomaly detection, and cross-cluster search are locked behind Platinum+ licensing. The enterprise bill is opaque and non-negotiable.
Ecosystem tax. Running ES means also running Kibana, Logstash, and Beats — services to operate before your first query.

THE XERJ ANSWER

Drop-in API compatibility. Port 9200, same endpoints: POST /:index/_search, POST /_bulk, POST /_aliases. Existing ES client libraries work unchanged.
No JVM, no GC pauses. No heap to size, no stop-the-world stalls. Predictable latency at every percentile, at a fraction of the memory footprint.
Smaller on disk. Domain-aware columnar compression with automatic encodings — 672.5 MB vs 806.7 MB on the same corpus (1.20× smaller).
One binary replaces five. No Kibana, no Logstash, no Beats, no separate vector DB. REST API + Prometheus metrics + native ingest — all in-process.
All features, one tier. Auth, TLS, RBAC, aggregations, vector search, hybrid queries — everything ships in the binary. No platinum paywall.

REPRODUCIBLE · ES 8.13 · MIXED QUERY TYPES · SCORECARD.md

XERJ ES 8.13

Bool query 1.38 ms 15.90 ms 11.5×

Query-string query 1.35 ms 9.35 ms 6.9×

Range filter 1.34 ms 4.66 ms 3.5×

kNN · k=10 · 100% recall 0.78 ms 2.64 ms 3.4×

Match query 0.57 ms 1.45 ms 2.5×

Term query 1.36 ms 2.15 ms 1.6×

P50 · 1M-DOC LLM-TELEMETRY CORPUS · FULL 91-CELL MATRIX AT xerj.org/benchmarks →

READY?·REQUEST ACCESS

RUN IT ON
YOUR DATA.

Send us your _mapping and a week of _bulk traffic. We'll run the migration on our infra and send you the head-to-head — query latency, ingest throughput, disk usage, memory — with reproduction scripts.

ELASTICSEARCHREPLACEMENT.

THE COST OF ELASTICSEARCH

THE XERJ ANSWER

SEE IT LIVE.

RUN IT ONYOUR DATA.

ELASTICSEARCH
REPLACEMENT.

RUN IT ON
YOUR DATA.