[Question]: if Simple Vector Store has no capacity setting

[waveBoom]

Question Validation

• I have searched both the documentation and discord for an answer.

Question

I want to know if ''Simple Vector Store'' has no capacity setting, that is, it will grow infinitely with the project (or it is only designed for experimentation or testing)

[dosubot]

The "Simple Vector Store" is designed without a built-in capacity setting, meaning it's intended to grow as needed without inherent limits, aside from those imposed by the system's available memory. This behavior is due to its storage mechanism, which utilizes a Python dictionary for storing embeddings, and Python dictionaries naturally expand based on the amount of stored data. Given this design choice, the "Simple Vector Store" is likely more suitable for experimentation or testing rather than for production environments where memory management and capacity limits are crucial.

For scenarios requiring a vector store with explicit capacity management, the LlamaIndex project offers alternatives that might be better suited for production use. These include options like Apache Cassandra®, Astra DB, and Azure AI Search, among others, which are designed to handle larger datasets with more sophisticated data management capabilities.

There's no current information available on whether there are plans to introduce capacity management features to the "Simple Vector Store" or to develop a new vector store with such features in future updates of LlamaIndex. For the most up-to-date information on this and other features, reviewing the project's documentation and source code is recommended.

Details

Sources

• llama-index-core/llama_index/core/vector_stores/simple.py
• benchmarks/vector_stores/bench_simple_vector_store.py
• docs/docs/module_guides/storing/vector_stores.md

About Dosu
This response is meant to be useful and save you time. It is not meant to be a precise solution, but rather a starting point for your own research.

Help me be more useful! Please leave a 👍 if this is helpful and 👎 if it is irrelevant.

To continue the conversation, mention @dosu.

[logan-markewich]

The "capacity" is the amount of memory you have, since all the vectors are stored in memory

It's really only meant to get people quickly started or work with smaller amounts of data. Actual use-cases that have larger data requirements etc. should use a vector index integration (qdrant, chroma, weaviate, etc.)

[waveBoom]

The "capacity" is the amount of memory you have, since all the vectors are stored in memory

It's really only meant to get people quickly started or work with smaller amounts of data. Actual use-cases that have larger data requirements etc. should use a vector index integration (qdrant, chroma, weaviate, etc.)

i get it , thanks