Engine Compatibility
Headband supports Meilisearch, Typesense, and Elasticsearch backends. The API is engine-agnostic -- the same endpoints work regardless of which engine your project uses.
When you create a project, you select the engine type and provide a connection URL. All subsequent API calls are translated to the appropriate engine protocol automatically.
The Elasticsearch engine is also compatible with OpenSearch. Point your connection at any OpenSearch node and Headband will work the same way.
Engine and machine advisor
The project creation flow includes a workload advisor for corpus size, traffic, and use case. Start with Meilisearch for small catalogs and docs, Typesense for autocomplete-heavy UI, and Elasticsearch/OpenSearch for logs, very large corpora, large multi-tenant content, or advanced RAG/vector workloads. Each recommendation now includes a plan fit, deployment path, ingestion plan, and upgrade trigger so teams can move from prototype to production deliberately.
Starter
Free/Pro fit: 1-2 vCPU / 1-4 GB RAM for dev, beta, and small production indexes. Direct writes are fine; backfill in small chunks.
Growth
Pro/Scale fit: 4-8 vCPU / 8-16 GB RAM when facets, filters, or autocomplete traffic rise. Add queues and replayable sync jobs.
Scale
Scale/BYOE fit: managed clusters with replicas for 1M+ docs, high QPS, or log data. Benchmark before committing SLOs.
When to upgrade
Move up when p95 search latency crosses 200-250ms, autocomplete exceeds 100ms, indexing competes with live search, reindex windows exceed your rollout budget, or restore/snapshot time exceeds your recovery objective.
Sizing guidance is a launch baseline, not a capacity guarantee. Benchmark real documents, filters, and p95/p99 latency targets before production rollout.
Production sizing runbook
1. Pick the retrieval job
Catalog, autocomplete, docs/RAG, logs, or multi-tenant content. Do not choose a machine before the workload is named.
2. Start with a tier
Starter: 1-2 vCPU. Growth: 4 vCPU / 8 GB. Performance: 8+ vCPU / replicas. Scale: benchmarked BYOE cluster.
3. Validate with traffic
Test filters, facets, sort, pagination, p95/p99 latency, zero-result rate, and A/B/C variants using real documents.
4. Package BataDB search
BataDB owns billing and add-on UX; Headband owns retrieval execution, index routing, API keys, and analytics.
Repo runbook: docs/engine-machine-sizing-runbook.md keeps the same operator checklist available outside the app.
Feature Matrix
CORS: The /v1 API supports CORS for browser-based usage. Search keys are safe to use in frontend applications. Admin keys should only be used server-side.