Vector Retrieval Service for Milvus:FAQ

How do I estimate the required computing resources?

You can use the resource calculator provided by Milvus to estimate the required resources. Before you deploy your service, adjust the resources based on test results. For more information, see Estimate required compute resources with resource calculator.

How do I choose between the Basic and Standard editions of Milvus?

Consider your scenarios and requirements:

• Basic Edition (single-node): This edition is ideal for testing or for users new to Milvus. It lets you quickly explore Milvus features.

• Standard Edition (cluster): This edition is designed for production environments and offers higher stability. Choose the Standard Edition for the best service experience and performance.

During the public preview, neither the Basic Edition nor the Standard Edition is covered by a Service-Level Agreement (SLA).

How do I configure alert rules?

To configure alert monitoring, go to the Monitoring and Alerting page in the Milvus console, select the Alert Settings tab, and enable the One-Click Alert feature. You can also click Alert Rule Settings to go to the CloudMonitor console and customize alert rules.

How do I choose an index?

Indexes are critical for optimizing vector search performance. Different index types have different impacts on query speed, precision, and resource usage. The following is a brief comparison of common Milvus index types and recommendations for choosing one:

• IVF_FLAT: A balanced choice for most scenarios.

  • How it works: Uses quantization to reduce computational complexity, significantly improving query performance over the FLAT index.

  • Best for: Scenarios with large datasets where you need a good balance between query performance, accuracy, and resource cost.

• HNSW (Hierarchical Navigable Small World): The ultimate choice for speed and accuracy.

  • How it works: Builds a multi-layered graph structure to efficiently navigate the vector space, enabling extremely fast and accurate searches, especially in high-dimensional data.

  • Best for: Scenarios that demand the highest query speed and recall, have ample memory and computing resources, and often involve high-dimensional data.

• IVF_SQ8: The high-density index for memory-constrained environments.

  • How it works: Uses Scalar Quantization (SQ8) to compress each vector, allowing you to fit significantly more vectors into the same amount of RAM. While this reduces memory costs, the quantization step can lead to a slight decrease in precision compared to non-compressed indexes like HNSW.

  • Best for: Scenarios where dataset size is very large and memory resources are the primary bottleneck. It's the go-to choice when you need to maintain a high recall rate on a budget and can tolerate a minor drop in accuracy.

• FLAT (exact index): The gold standard for 100% accuracy.

  • How it works: Performs an exhaustive, brute-force search by comparing the query vector against every single vector in the index. This is the only method that guarantees finding the true, exact nearest neighbors.

  • Best for: Applications where 100% recall is non-negotiable, dataset sizes are relatively small (e.g., under a few million), and query latency is not a primary concern.

How do I change the configuration of an instance?

You can query and modify instance configurations in the Milvus console to meet your business requirements. For more information, see Update instance configurations.

How do I reset the instance password?

You can reset the password for a Milvus instance using one of the following methods.

from pymilvus import MilvusClient

# Create a Milvus client.
client = MilvusClient(
    uri="http://c-b26606e936****.milvus.aliyuncs.com:19530",  # The public endpoint of the Milvus instance.
    token="<yourUsername>:<yourPassword>",  # The username and password to log on to the Milvus instance.
    db_name="default"  # The name of the database to connect to. This example uses the default database.
)

client.update_password(
    user_name="root",  # Specify the username whose password you want to change. This example uses root.
    old_password="old_password",  # Enter the old password.
    new_password="new_password",  # Enter the new password.
    using="default"  # The database name.
)

Why can't I access my Milvus instance?

If you cannot access your Milvus instance, follow these steps to troubleshoot and resolve the issue:

1. Check network connectivity.

   From your command line, run the following command to check the network connection to your Milvus instance. Replace c-xxxx.milvus.aliyuncs.com with your actual public domain name.

   telnet c-xxxx.milvus.aliyuncs.com 19530

   • If the output is Connected to c-xxxx.milvus.aliyuncs.com, the network connection is normal.

   • If the connection fails, confirm that Internet access is enabled and that your IP address is added to the instance's public access whitelist.

2. Check Python SDK compatibility.

   After you confirm that the network is connected, run pip list on your local command line to check whether your Python SDK version meets the requirements. If your pymilvus version is earlier than 2.0, upgrade it and try to connect again.

How do I use the Python SDK to access an Milvus instance?

You can install the PyMilvus library locally to connect to your Milvus instance. For more information, see Quickly implement vector search.

How do I grant Milvus permissions to a RAM user?

You can use your account to grant the `AliyunMilvusFullAccess` permission to a Resource Access Management (RAM) user. For more information, see Authorize a RAM user.

Does Milvus support graphical management?

Yes, it does. Milvus integrates Attu for more efficient management of databases, collections, indexes, and entities. For more information, see Manage tools with Attu.

How do I access the Attu Manager page?

You must enable Internet access for the instance and add your local public IP address to the whitelist. For more information, see Access the Attu page.