Think Build Implement Repeat
London, UK +44 7367 067226
WhatsApp FOLLOW f in X
Python & Django

Vector Storage for a Business System

Last updated:

Start with what you already run

Most business retrieval systems involve tens of thousands of chunks, not hundreds of millions. At that scale, a vector extension to your existing database performs perfectly well.

Adding a dedicated vector database to a system with forty thousand chunks is adding an operational dependency for no measurable benefit.

The options

OptionSuitsOperational cost
Database extensionUp to millions of chunksNone — already running it
Managed vector serviceLarger scale, no ops capabilitySubscription
Self-hosted vector databaseVery large scale, specific needsReal
In-memory indexSmall, static datasetsLow, but not durable

What actually matters

  1. Metadata filtering during search, not after — essential for permissions
  2. Update performance, since documents change
  3. Backup and recovery, like any other data store
  4. Operational familiarity within your team

Filtering during search is the requirement people discover late, usually when implementing permissions.

Keep the source of truth elsewhere

The vector store is an index, not a system of record. Documents should live in their proper store, with the index rebuildable from them.

That means a corrupted or lost index is an inconvenience rather than data loss, and it makes changing vector store later straightforward.

Plan for reindexing

  • Changing embedding model requires a full reindex
  • Changing chunking strategy requires a full reindex
  • Both will happen at some point
  • Make it a scripted operation rather than a project

Frequently asked questions

When is a dedicated vector database worth it?

At scale where your existing database struggles, or where you need features it does not provide. Below that it is complexity without benefit.

How much storage do embeddings need?

Modest for typical business volumes. Tens of thousands of chunks is a few hundred megabytes, not a capacity problem.

Can we change vector store later?

Yes, if the source documents remain the system of record and reindexing is scripted. Design for that from the start.

What about hybrid search?

Combining keyword and semantic search improves results, particularly for identifiers. Check your chosen store supports it.

Keep reading

Being told you need a vector database?

At most business volumes, an extension to your existing database is enough. Happy to look at your numbers.

Book a free 30-minute call Get a project estimate WhatsApp us

Related services

What we build for problems like this one

Custom Software DevelopmentWeb DevelopmentMachine Learning