Milvus Monitoring

Danger

Milvus Monitoring Service End-of-Service Notice

The Milvus Monitoring service currently provided will end on December 31, 2026. Accordingly, existing Milvus Monitoring projects will be available until December 31, 2026, and metric collection through these projects will end starting January 1, 2027.

Afterward, you can switch to OpenMetrics. The migration may require installing an agent or changing configurations, and the collection method and billing criteria may change.

Note

Feature Project

WhaTap newly introduced Feature Project (Features) so that users can easily check whether the desired software supports monitoring and easily start monitoring. The Feature Project provides an optimized monitoring solution by repackaging existing monitoring products to fit the software to monitor. Manage your services for stability and performance more effectively through the WhaTap's Feature Project.

The Milvus open source database is used for high-performance management and search of large vector data. The WhaTap Milvus monitoring traces the performance and resource usage for Proxy, Root Coordinator, Query Node, and such that make up the Milvus cluster in real time and historically, helps you quickly discover and resolve search delays, memory shortages, and transaction failures.

WhaTap provides *** Milvus*** monitoring in a new feature project manner. By tracing the performance and status of Milvus database in real time, it helps you identify and resolve problems early as possible.

Installation

This guide provides the basic installation instructions for using the WhaTap Milvus monitoring service.

To use the WhaTap monitoring service, after Sign up, create a project and then install the agent to the target server. For more information about membership registration, see the following.

Note

Supported environment

Before installing the WhaTap *** Milvus*** monitoring agent, check the support environment.

• Milvus: Milvus 2.x or equivalent

• Ubuntu: Ubuntu 12.04 or later

• OS: Red Hat 6 or equivalent (CentOS, Rocky Linux, Amazon Linux)

• OS Architecture: Amd64/X86_64, Arm64/Aarch64

Creating a Feature Project

Create a project before installing the agent.

1. Log in to the WhaTap Monitoring Service.

2. In the left side menu, click All Projects > + Project.

3. On the Select product screen, select the product you want to install.

4. Enter or select the following items:

   • Project name: Enter the project name.

   • Data server region: Select the data server region. A region is a group of data centers that provide cloud services. Your data will be stored in the selected region.

   • Time zone: Set the time zone to be used for alerts and report generation.

   • Notification language setting: Set the language for alert notifications. (Supports Korean and English)

   • Project groups: Group multiple projects for easier management. Select a group if applicable.

   • Project description: Add additional details or a description for your project.

5. After completing all settings, click the Creating a project button.

Note

If you add a project while an organization is selected, you must set the Groups of organization field.

For more information about groups, see the related document.

Milvus agent installation

After creating a Feature Project (Features), the Install Milvus Agent screen appears. Proceed with installation according to the following instructions.

1. Check the project access key.

   Project access key is the unique ID for activating the WhaTap services. Select Getting the access key.

2. Create the installation script.

   Execute the following command to automatically have the system detect the user environment and generate an installation script for the server with Milvus installed.

   curl https://repo.whatap.io/telegraf/feature/milvus/install_milvus_monitoring.sh -o install_milvus_monitoring.sh

3. After configuring and restarting the agent, start monitoring.

   Run the following command to install the WhaTap Server Monitoring Agent and Telegraph Sidecar.

   chmod +x install_milvus_monitoring.sh
   sudo ./install_milvus_monitoring.sh "x434522kkhbtj-x5cv0fliua45of-zmgqva9eakmg9" "15.165.146.117"

Learn about the main features

Dashboard

Select Home screen > Feature project(Features) > *** Milvus*** > *** Milvus Dashboard***

The WhaTap Milvus monitoring provides key metrics and graphs by each component of the Milvus that is the vector search database. The ** Milvus Dashboard** provided by WhaTap consists of the following three presets:

• Milvus Proxy and Root Coordinator

  This preset provides comprehensive monitoring of Proxy latency, QPS, Root Coordinator's collections and entities, and DML/DDL operation status in the Milvus system.

  Guide to dashboard metrics

  Proxy Monitoring

  • Latency and request events

    You can visualize the latency of requests processed by the proxy as a histogram and check the number of requests per second (QPS), and the like.

  • Process status

    By displaying the CPU and memory usage of Milvus Proxy in a time series graph, the status of system resources can be checked.

  Root Coordinator Monitoring

  • Collection and entity information

    Check the database size through the number of collections and entities in operation.

  • DML and DDL operations

    By visualizing the DML operations such as Insert and Delete and DDL operations such as Create and Drop, you can check the frequency of operations.

  • Streams and Message Channels

    By tracing the numbers of data streams and message channels, you can check the internal data flow within Milvus.

• Milvus Search and Compute Performance

  This preset monitors the search performance such as ANN, Brute Force, and DiskANN.

  Guide to dashboard metrics

  Approximate Nearest Neighbor (ANN) Metrics

  • Search request delay time

    You can analyze the search performance by checking the delay time of ANN search in the form of histogram.

  • Number of search requests (QPS) and total latency

    You can monitor the system load by checking the number of search requests processed per second and cumulative search latency.

  BF (Brute Force) Search

  • Number of objects compared per BF search request

    Check the number of objects compared per request through the Brute Force search.

  • Total BF computation amount

    You can trace computational costs and find optimization opportunities through the number of object comparisons.

  Bitset Ratio

  • Bitset utilization and calculation amount

    You can analyze opportunities to improve the search speed by monitoring whether operations are optimized using Bitset.

  • Number of Bitset application requests and number of objects applied

    You can evaluate the effectiveness of the indexing strategy by checking the percentage of search requests that use Bitset.

  DiskANN

  • DiskANN search and index performance

    You can analyze the performance for disk-based searches by tracing the latency and search count of search requests using DiskANN.

  • DiskANN index size and number of search requests

    You can optimize the efficiency by comparing the space occupied by the disk index with the number of search requests.

• Milvus Query Coordinator & Query Node Metrics

  This preset monitors the status and performance metrics for QueryCoord and QueryNode in real time.

  Guide to dashboard metrics

  Query Coordinator

  • Number of collections

    It indicates the number of collections managed by QueryCoord.

  • Number of partitions

    By monitoring the number of partitions within a collection, you can check the data partitioning status.

  • Number of QueryNodes

    It displays the number of QueryNodes managed by QueryCoord.

  • Resource Group Info

    It analyzes query execution resources by checking the resource group allocated in QueryCoord.

  Query Node

  • Disk used size(MB)

    It monitors the disk space used by QueryNode to optimize the storage.

  • Number of ready read tasks in readyQueue

    It monitors the number of pending query requests to analyze whether there are any search bottlenecks.

  Data Coordinator & Data Node

  • Number of collections

    It traces data size by displaying the number of collections managed by DataCoord.

  • Number of data nodes

    You can check the storage management status through the number of DataNodes managed by DataCoord.

  • Import tasks grouping by type and state

    You can analyze the loading and process speeds by classifying data import tasks by type and status.