Diese Seite ist in Ihrer lokalen Sprache noch nicht verfügbar. Wir arbeiten daran, weitere Sprachversionen hinzuzufügen. Vielen Dank für Ihre Unterstützung.
- What's New
- Function Overview
- Service Overview
- Getting Started
-
User Guide
- Getting Started
- Creating and Accessing a Cluster
- Scaling In/Out a Cluster
- Upgrading Versions
- Importing Data to Elasticsearch
-
Managing Elasticsearch Clusters
- Cluster and Storage Capacity Statuses
- Introduction to the Cluster List
- Index Backup and Restoration
- Binding an Enterprise Project
- Restarting a Cluster
- Migrating Cluster Data
- Deleting a Cluster
- Managing Tags
- Public Network Access
- Managing Logs
- Managing Plugins
- Hot and Cold Data Storage
- Configuring Parameters
- VPC Endpoint Service
- Kibana Public Access
- Vector Retrieval
- Working with Kibana
- Elasticsearch SQL
- Connecting a Cluster to a Dedicated Load Balancer
- Enhanced Features
- Monitoring
- Auditing
- Change History
- Best Practices
-
API Reference
- Before You Start
- API Overview
- Calling APIs
- Getting Started
-
Cluster Management
- Creating a cluster
- Querying the Cluster List
- Querying Cluster Details
- Deleting a Cluster
- Renaming a Cluster
- Changing the Password of a Cluster
- Restarting a Cluster
- Scaling Out a Cluster
- Adding Instances and Expanding Instance Storage Capacity
- Changing Specifications
- Obtaining the Instance Specifications List
- Querying All Tags
- Querying Tags of a Specified Cluster
- Adding Tags to a Cluster
- Deleting a Cluster Tag
- Adding or Deleting Cluster Tags in Batches
- Changing the Specifications of a Specified Node Type
- Scaling In a Cluster by Removing a Specific Node
- Scaling In Nodes of a Specific Type
- Downloading a Security Certificate
- Replacing a Node
- Configuring the Security Mode.
- Adding Independent Masters and Clients
- Changing the Security Group
- Kibana Public Network Access
- Log Management
- Public Network Access
-
Snapshot Management
- (Not Recommended) Automatically Setting Basic Configurations of a Cluster Snapshot
- Modifying Basic Configurations of a Cluster Snapshot
- Manually Creating a Snapshot
- Restoring a Snapshot
- Deleting a Snapshot
- Configuring the Automatic Snapshot Creation Policy
- Querying the Automatic Snapshot Creation Policy
- Querying a Snapshot List
- Disabling the Snapshot Function
- VPC Endpoint
- Parameter Configuration
- Common Parameters
- Change History
- SDK Reference
-
FAQs
- General Consulting
-
Accessing CSS Clusters
- How Do I Reset the Administrator Password of a Security-mode Cluster in CSS?
- Are Ports 9200 and 9300 Open for Access to Elasticsearch Clusters?
- How Do I Use a NAT Gateway to Access CSS from the Internet?
- How Do I Connect In-house Developed Kibana to an Elasticsearch Cluster in CSS?
- How Do I Connect In-house Developed OpenSearch Dashboards to an OpenSearch Cluster in CSS?
- Migrating CSS Clusters
-
Using CSS Cluster Search Engines
- Why Are Newly Created Index Shards Allocated to a Single Node in CSS?
- How Do I Create a Type Under an Index in an Elasticsearch 7.x Cluster of CSS?
- How Do I Change the Number of Replicas for Elasticsearch Indexes in CSS?
- What Are the Impacts If an Elasticsearch Cluster of CSS Has Too Many Shards?
- How Do I Check the Number of Shards and Replicas in a CSS Cluster?
- What Does the Value i for node.roles Mean for Nodes in an Elasticsearch Cluster of CSS?
- How Do I Change the Maximum Number of Results Returned for Searches to an Index in an Elasticsearch Cluster of CSS?
- How Do I Update Index Lifecycle Policies for an Elasticsearch Cluster of CSS?
- How Do I Set Slow Query Log Thresholds for an Elasticsearch Cluster of CSS?
- How Do I Clear Elasticsearch Indexes in CSS?
- How Do I Clear Elasticsearch Cache in CSS?
- Why Does the Disk Usage Increase After the delete_by_query Command Was Executed to Delete Data in an Elasticsearch Cluster?
- Do CSS Elasticsearch Clusters Support script dotProduct?
-
Managing CSS Clusters
- How Do I Check the AZ Where a CSS Cluster Is Located?
- What Is the Relationship Between the Filebeat Version and Cluster Version in CSS?
- How Do I Obtain the Security Certificate of CSS?
- How Do I Convert the Format of a CER Security Certificate in CSS?
- Can I Modify the Security Group for Elasticsearch and OpenSearch Clusters in CSS?
- How Do I Set search.max_buckets for an Elasticsearch Cluster of CSS?
- Can I Modify the TLS Algorithm of an Elasticsearch or OpenSearch Cluster in CSS?
- How Do I Enable Audit Logs for an Elasticsearch or OpenSearch Cluster of CSS?
- Can I Stop a CSS Cluster?
- How Do I Query the Index Size on OBS After the Freezing of Indexes for a CSS Cluster?
- How Do I Check the List of Default Plugins for Elasticsearch and OpenSearch Clusters?
- CSS Cluster Backup and Restoration
- CSS Cluster Monitoring and O&M
- Troubleshooting
- Videos
Show all
Description
Large-scale image recognition and retrieval, video search, and personalized recommendation impose high requirements on the latency and accuracy of high-dimensional space vector retrieval. To facilitate large-scale vector search, CSS integrates the vector search feature powered by Huawei's vector search engine and the Elasticsearch plug-in mechanism.
Principles
Vector search works in a way similar to traditional search. To improve vector search performance, we need to:
- Narrow down the matched scope
Similar to traditional text search, vector search use indexes to accelerate the search instead of going through all data. Traditional text search uses inverted indexes to filter out irrelevant documents, whereas vector search creates indexes for vectors to bypass irrelevant vectors, narrowing down the search scope.
- Reduce the complexity of calculating a single vector
The vector search method can quantize and approximate high dimensional vectors first. By doing this, you can acquire a smaller and more relevant data set. Then more sophisticated algorithms are applied to this smaller data set to perform computation and sorting. This way, complex computation is performed on only part of the vectors, and efficiency is improved.
Vector search means to retrieve the k-nearest neighbors (KNN) to the query vector in a given vector data set by using a specific measurement method. Generally, we only focus on Approximate Nearest Neighbor (ANN), because a KNN search requires excessive computational resources.
Functions
Our engine integrates a variety of vector indexes, such as brute-force search, Hierarchical Navigable Small World (HNSW) graphs, product quantization, and IVF-HNSW. It also supports multiple similarity calculation methods, such as Euclidean, inner product, cosine, and Hamming. The recall rate and retrieval performance of the engine are better than those of open-source engines. It can meet the requirements for high performance, high precision, low costs, and multi-modal computation.
The search engine also supports all the capabilities of the native Elasticsearch, including distribution, multi-replica, error recovery, snapshot, and permission control. The engine is compatible with the native Elasticsearch ecosystem, including the cluster monitoring tool Cerebro, the visualization tool Kibana, and the real-time data ingestion tool Logstash. Several client languages, such as Python, Java, Go, and C++, are supported.
Constraints
- Only clusters of version 7.6.2 and 7.10.2 support vector search.
- The vector search plug-in performs in-memory computing and requires more memory than common indexes do. You are advised to use memory-optimized computing specifications.
Feedback
Was this page helpful?
Provide feedbackThank you very much for your feedback. We will continue working to improve the documentation.