FAQ: Get Started
This topic lists the possible issues that you may encounter while you get started with Zilliz Cloud and the corresponding solution.
Contents
- Is there any performance comparison between Zilliz Cloud and other vector search solutions?
- Which type of index is supported by Zilliz Cloud?
- What is the search latency of Zilliz Cloud?
- How can I get further technical support?
FAQs
Is there any performance comparison between Zilliz Cloud and other vector search solutions?
Yes. You can use VectorDBBench, a vector database benchmark tool to compare the performance of Zilliz Cloud and other mainstream vector databases and cloud services.
Which type of index is supported by Zilliz Cloud?
Currently, Zilliz Cloud only supports AUTOINDEX, a proprietary index type that can help you achieve better search performance.
For a performance-optimized cluster with 1.5 million 768-dimensional vectors, the QPS can reach several hundred and the latency is below 100 milliseconds. For a capacity-optimized cluster with 5 million 768-dimensional vectors, the QPS can reach 50 and the latency is over 200 milliseconds. For more details, see AUTOINDEX Explained.
However, please submit a request if you are familiar with using any of the indexes we support. We can help you evaluate your application demand and enable the indexes for you.
What is the search latency of Zilliz Cloud?
The search latency depends on the CU type and data volume.
top_k | Latency of Performance-optimized CU (768-dim 1M vectors) | Latency of Capacity-optimized CU (768-dim 5M vectors) |
---|---|---|
10 | < 10 ms | < 50 ms |
100 | < 10 ms | < 50 ms |
250 | < 10 ms | < 50 ms |
1000 | 10 - 20 ms | 50 - 100 ms |
For more details about the test result, see Select the Right CU.
How can I get further technical support?
Please submit at request at the Zilliz cloud support portal.