Milvus Vector Database, Zilliz Cloud (Pay-as-you-go)

I am primarily using Zilliz Cloud for a managed vector database, storage, and searching and indexing.

My primary workload involves storing and searching high-dimensional vector embeddings that are generated from documents and all the knowledge bases that I have, along with all the technical contents and the application data that I have. This platform is a core component of my RAG system architecture. Prior to adopting this managed vector database, I also tried self-hosted Milvus, which is not very scalable and has very high setup overhead, and so we decided to use Zilliz Cloud.

Zilliz Cloud has allowed us to focus on building the AI products without the overhead of operating vector database infrastructure.

The best features Zilliz Cloud offers, in my opinion, include high-performance similarity search, managed infrastructure for cluster maintaining, infrastructure scaling, backup management, storage planning, Milvus compatibility, and metadata filtering.

The offering of this managed infrastructure of a vector database is most useful for me, and the high-performance similarity search is useful in my case.

Regarding the similarity search, it delivers low latency retrieval and maintains strong relevance in returned units, which is particularly useful for me.

Zilliz Cloud has positively impacted my organization because initially, we spent too much time hosting self-hosted Milvus and planning for infrastructure that did not yield very useful results. Now, we do not have the overhead of managing infrastructure for my vector database, so we can directly focus on building our RAG system and AI workload.

Time saved is the first and foremost outcome since all the time we invested in self-hosting Milvus has been redirected towards building the AI workloads. Time has definitely been saved, which is the primary benefit of using Zilliz Cloud.

It can be improved a little bit on the search functionality.

Not in specific search functionality, but I would like to see more visibility in the costing part and the monitoring dashboard.

Zilliz Cloud could provide more automated optimization guidance, particularly for large-scale deployments, mostly around index selection, partitioning, and resource sizing, which would help maximize performance.

I have been using Zilliz Cloud for the past one year.

If you are building some AI workloads, do not focus on managing the vector database. It is an operational overhead. Start building the AI workload instead. I would rate this product a 9 out of 10.

Public Cloud

Amazon Web Services (AWS)

I have Milvus hosted on Zilliz Cloud and am majorly using it to manage the vector database and utilizing some of the RAG and vector features from that.

I connected Zilliz Cloud with a lot of Kubernetes clusters on Zilliz Cloud to fetch a lot of data because we have a client who wants to use CNN models to give the best result from their database. We have RAG, which is using vector embedding, and we manage everything on AWS, where some of the services are connected with Zilliz Cloud to gather everything faster.

The best feature of Zilliz Cloud is that it helps in very high-performance vector search, and it is also very scalable, with very low latency that helps provide faster results. The deployment of Milvus is very easy because it is managed there, so I did not need to take care of anything. These are the major things that I feel are very important.

Zilliz Cloud has positively impacted my organization because previously, when I was not using it, there was a little lag in the output of the search due to the lack of a proper vector search setup, and maintaining the vector search was very hard, requiring me to create a model, deploy it, and connect everything. It helped me a lot by using managed Milvus, which simplifies my management tasks.

Having more connections with all other major clouds could be helpful, and a marketplace could grow with Zilliz Cloud.

I have been using Zilliz Cloud for around seven to eight months.

Zilliz Cloud is stable in my experience.

Its scalability is very good.

The customer support is also good.

I have not used any different solutions before Zilliz Cloud.

The biggest return on investment I have seen is in the time saved in my current scenario.

The pricing, setup cost, and licensing experience were pretty straightforward, and although I was not involved with the team, I felt it was smooth.

Before choosing Zilliz Cloud, I evaluated Weaviate and PineconeDB.

If others do not have the bandwidth to manage the vector search and maintain that on the cloud, I recommend that they find it very easy to use Zilliz Cloud.

Zilliz Cloud is deployed in my organization on a public cloud.

I use AWS as my cloud provider.

I did not purchase Zilliz Cloud through the AWS Marketplace; the company directly purchased it.

Zilliz Cloud helps a lot, and I also contribute to the community while creating a lot of awareness for people to use it. I would rate this review an overall eight out of ten.

Public Cloud

Amazon Web Services (AWS)

What do you like best about the product?

Profesional technical support, great documentation and the most scalable and reliable solution for vector similarity search

What do you dislike about the product?

Nothing to dislike. Everything is going great so far

What problems is the product solving and how is that benefiting you?

Zilliz addresses the challenge of searching through embeddings created from social media creators' content on a large scale. It effectively manages the complexity involved in handling and retrieving relevant information from vast amounts of such data.

What do you like best about the product?

Zilliz has been pretty stable in the last year. The team was very helpful resolving issue in the initial integration period. We currently store about 50 million vectors.

What do you dislike about the product?

The cost can be somewhat high when we stored like 1 billion vectors.

What problems is the product solving and how is that benefiting you?

We use Zilliz as our RAG

What do you like best about the product?

This product is quick, affordable, straightforward, and gets the job done. It's user-friendly and can easily scale to meet growing needs.

What do you dislike about the product?

Modifying collection schemas involves a migration process, which can be quite complex and often takes a significant amount of time to manage.

What problems is the product solving and how is that benefiting you?

Zilliz serves as the primary knowledge base for our AI agent. It drives the RAG functionality for a customer support and shopping assistant AI agent that currently assists millions of e-commerce shoppers.

What do you like best about the product?

Extremely fast, low-latency vector search, even at a massive scale. The cloud platform is intuitive, and the SDKs (like PyMilvus) are straightforward to integrate.

What do you dislike about the product?

The pricing for the cloud service can be a bit high for smaller projects or individual developers.

What problems is the product solving and how is that benefiting you?

We use Zilliz to power our recommendation system. It solves the critical problem of storing and searching millions of image embeddings at high speed. This benefits us by enabling real-time, accurate visual recommendations and 'similar item' search for our users, which significantly improves user engagement.

What do you like best about the product?

I truly love using Zilliz for building our video retrieval tool as it has greatly enhanced our ability to convert each frame of a video into vectors and store them effectively in Zilliz Cloud. This functionality allows for accurate and efficient searches for corresponding images or videos, whether making searches from text to images or vice versa. The support from Zilliz is impeccable, with a very active community and extremely reliable database systems, which gives us a great deal of confidence. The most impressive aspect is the retrieval speed; it is exceptionally fast, allowing us to search through millions of vectors in just 200 milliseconds, which is critical for our operations. Additionally, the ability of Zilliz to handle a large capacity of data further boosts its value, making our data management and retrieval processes smooth and efficient.

What do you dislike about the product?

I think the number of vector columns supported in a single collection is a bit low, it seems to be 4 or 8. We would like to be able to search across more dimensions.

What problems is the product solving and how is that benefiting you?

I use Zilliz to convert video frames into vectors stored in Zilliz Cloud, facilitating accurate image or video retrieval. The retrieval speed is extremely fast, processing millions of vectors in 200 milliseconds, and the capacity is very large.

What do you like best about the product?

I really appreciate how Zilliz supports hybrid retrieval, combining both vector and keyword searches. This feature is incredibly beneficial as it provides me with more accurate, context-aware results and ensures that I do not miss important exact matches. It’s very convenient to use, making my workflow more efficient. Additionally, the initial setup was a breeze thanks to Docker Compose and the clear documentation which guided me quickly through getting started.

What do you dislike about the product?

I find that sometimes the documentation could be more detailed, especially for advanced configurations and performance tuning. They could include more real-world configuration examples, detailed parameter explanations, and best practices for optimizing performance with large-scale hybrid retrieval workloads.

What problems is the product solving and how is that benefiting you?

I use Zilliz for RAG, enhancing search accuracy and document storage with similarity search and hybrid retrieval, combining vector and keyword search for context-aware results.

What do you like best about the product?

Excellent performance, Automatic updates,Comprehensive functionality,Supports extremely large datasets

What do you dislike about the product?

Support for scalar fields is not perfect.
The price is relatively expensive.

What problems is the product solving and how is that benefiting you?

It serves as a vector data storage solution for RAG data retrieval in AI scenarios.
It is also the primary database for semantic similarity matching calculations.

What do you like best about the product?

I love how Zilliz excels at scalability and high-performance vector search, making it incredibly efficient for handling massive datasets in AI-driven tasks. In my experience, its hybrid retrieval capabilities—combining vector embeddings with knowledge graph integrations—stand out, especially for complex queries involving semantic understanding. It's easy to integrate with tools like LangChain and OpenAI embeddings, and the cloud-native features in Zilliz Cloud simplify management without sacrificing speed or accuracy.

What do you dislike about the product?

For self-hosted Milvus setups, the configuration might feel a bit involved compared to fully managed options, but Zilliz Cloud largely resolves that with its user-friendly UI and elastic scaling.

What problems is the product solving and how is that benefiting you?

Zilliz is tackling challenges in entity alignment and disambiguation within my knowledge graph project by leveraging vector embeddings and graph-based retrieval to link and resolve entities across diverse datasets, reducing ambiguities (like distinguishing similar terms in different contexts) and enabling precise multi-hop reasoning. This has streamlined entity matching, cutting down on manual corrections and improving overall graph accuracy. In the recall step of my RAG pipeline, it enhances retrieval of relevant documents through hybrid searches (vector similarity plus relational structures), minimizing hallucinations in LLM outputs and providing more complete, contextually rich responses. Overall, it's saved me significant time on data processing, boosted the reliability of my AI applications, and allowed for handling larger-scale data without performance drops—ultimately making my projects more efficient and effective.