Milvus VS ann-benchmarks

Zilliz (zilliz.com) | Hybrid/ONSITE (SF, NYC) | Full-time

I am part of the hiring team for DevRel

NYC - https://boards.greenhouse.io/zilliz/jobs/4307910005

SF - https://boards.greenhouse.io/zilliz/jobs/4317590005

Zilliz is the company behind Milvus (https://github.com/milvus-io/milvus), the most starred vector database on GitHub. Milvus is a distributed vector database that shines in 1B+ vector use cases. Examples include autonomous driving, e-commerce, and drug discovery. (and, of course, RAG)

We are also hiring for other roles that I am not personally involved in the hiring process for such as product managers, software engineers, and recruiters.

Get started with Milvus on GitHub.

Milvus: An open-source vector database designed for AI and ML applications. It excels in handling large-scale vector similarity searches, making it suitable for recommendation systems, image and video retrieval, and natural language processing tasks.

Selecting the right version of open-source Milvus is important to the success of any project leveraging vector search technology. With Milvus offering different versions of its vector database tailored to varying requirements, understanding the significance of selecting the correct version is key for achieving desired outcomes.

Milvus is an open-source vector database designed to handle large-scale similarity search and vector indexing. It supports multiple index types and offers highly efficient search capabilities, making it suitable for a wide range of AI and ML applications, including image and video recognition, natural language processing, and recommendation systems.

Zilliz is hiring! We're looking for REMOTE and/or HYBRID roles in SF

Zilliz is the company behind Milvus (https://github.com/milvus-io/milvus), the most widely adopted vector database. Vector databases are a crucial piece of any technology stack looking to take advantage of unstructured data. Most recently and notably, Retrieval Augmented Generation (RAG). For RAG, vector databases like Milvus are used as the tool to inject customized data. In other words, vector databases make things like customized chat bots, personalized product recommendations, and more possible.

We are hiring for Developer Advocates, Senior+ Level Engineers and Product people, and Talent Acquisition. Check out all the roles here: https://zilliz.com/careers

Good on them, I know the crustaceans are out here happy about this raise for a Rust based Vector DB!

(now I'm gonna plug what I work on)

If you're interested in a more scalable vector database written in Go, check out Milvus (https://github.com/milvus-io/milvus)

But before we do, I do want to say that 🤩 all these lovely Open-Source projects would love a little 🎉💕 love by getting a GitHub star ⭐ for their efforts. Including Open Source Milvus 🥰

1. Milvus by Zilliz | Github

On top of my head, pgvector only supports 2 indexes, those are running in memory only. They don't support GPU indexing, nor Disk based indexing, they also don't have separation of query and insertions.

Also with different people I've talked to, they struggle with scale past 100K-1M vector.

You can also have a look yourself from a performance perspective: https://ann-benchmarks.com/

https://ann-benchmarks.com/ is a good resource covering those libraries and much more.

We utilized the ANN Benchmarks methodology, a standard for benchmarking vector databases. Our tests used the dbpedia dataset of 1,000,000 OpenAI embeddings (1536 dimensions) and inner product distance metric for both Pinecone and pgvector.

Data warehouses are columnar stores. They are very different from row-oriented databases - like Postgres, MySQL. Operations on columns - e.g., aggregations (mean of a column) are very efficient.

Most vector databases use one of a few different vector indexing libraries - FAISS, hnswlib, and scann (google only) are popular. The newer vector dbs, like weaviate, have introduced their own indexes, but i haven't seen any performance difference -

Reference: https://ann-benchmarks.com/

(Blog author here). Thanks for the question. In this case the index for both DiskANN and pgvector HNSW is small enough to fit in memory on the machine (8GB RAM), so there's no need to touch the SSD. We plan to test on a config where the index size is larger than memory (we couldn't this time due to limitations in ANN benchmarks [0], the tool we use).

To your question about RAM usage, we provide a graph of index size. When enabling PQ, our new index is 10x smaller than pgvector HNSW. We don't have numbers for HNSWPQ in FAISS yet.

[0]: https://github.com/erikbern/ann-benchmarks/

In terms of "All You Need" for Vector Search, ANN Benchmarks (https://ann-benchmarks.com/) is a good site to review when deciding what you need. As with anything complex, there often isn't a universal solution.

txtai (https://github.com/neuml/txtai) can build indexes with Faiss, Hnswlib and Annoy. All 3 libraries have been around at least 4 years and are mature. txtai also supports storing metadata in SQLite, DuckDB and the next release will support any JSON-capable database supported by SQLAlchemy (Postgres, MariaDB/MySQL, etc).

pg_vector doesn't perform well compared to other methods, at least according to ANN-Benchmarks (https://ann-benchmarks.com/).

txtai is more than just a vector database. It also has a built-in graph component for topic modeling that utilizes the vector index to autogenerate relationships. It can store metadata in SQLite/DuckDB with support for other databases coming. It has support for running LLM prompts right with the data, similar to a stored procedure, through workflows. And it has built-in support for vectorizing data into vectors.

For vector databases that simply store vectors, I agree that it's nothing more than just a different index type.

The reason https://ann-benchmarks.com is so good, is that we can see a plot of recall vs latency. I can see you have some latency numbers in the leaderboard at the bottom, but it's very difficult to make a decision.

As a practitioner that works with vector databases every day, just latency is meaningless to me, because I need to know if it's fast AND accurate, and what the tradeoff is! You can't have it both ways. So it would be helpful if you showed plots showing this tradeoff, similar to ann-benchmarks.