High-Performance Graph Database + GNN Intelligent Computing Platforms Makes More Business Possibilities Achievable
Enterprise-Level One-Stop Graph Intelligence Platform
eZoo Graph CORE FEATURES
GDB
As the foundation of the eZoo Graph Platform, eZoo GDB is a native, ultra-high-performance graph database developed completely in-house. It is based on C++, and its kernel is implemented through various core technologies such as data compression, memory management, lock-free concurrency, etc. It provides basic capabilities such as data import, graph management, and graph query for upper-level applications, and has dozens of built-in graph algorithms. It is deeply integrated with eZoo GNN.
VISUALIZATION
Graph visualization provides complete capabilities based on graphs, including graph definition, graph management, and graph query. Users can perform operations such as data import and graph definition through an interactive interface. They can also perform operations on created graphs, such as querying nodes, edges, paths, neighbors, etc., and map them onto a visual interface.
GNN
Integrating GNN computing capabilities, graph neural network computations can be performed based on GDB.
This supports graph data conversion for GNN, graph neural network training and inference, and graph model management.
eZoo STUDIO
A full set of graphical user interfaces for graph operations is provided
Built on the eZoo ultra-high-performance graph database (eZoo GDB), a full set of graphical user interfaces is provided for easy and quick querying of various graph operations such as neighbors, paths, subgraphs, etc.
A full set of graphical user interfaces for graph operations is provided
Built on the eZoo ultra-high-performance graph database (eZoo GDB), a full set of graphical user interfaces is provided for easy and quick querying of various graph operations such as neighbors, paths, subgraphs, etc.
Multiple graph algorithms and visualization tools
Built-in multiple graph algorithms and graph visualization tools, helping to mine more value. They include centrality, similarity, community detection, path finding, node mapping, topology link prediction, etc. There are dozens of built-in graph algorithms and graph visualization tools, which help to mine more value.
Graph Neural Network (GNN)
From data management to intelligent analysis, the eZoo graph platform is deeply integrated with GNN. As an integrated graph intelligence platform, eZoo deeply integrates Graph Neural Networks (GNN) into the entire AI data processing workflow, from data import, cleaning, exploration, model training, to online real-time inference, supporting the complete AI data processing process.
Graph Neural Network (GNN)
From data management to intelligent analysis, the eZoo graph platform is deeply integrated with GNN. As an integrated graph intelligence platform, eZoo deeply integrates Graph Neural Networks (GNN) into the entire AI data processing workflow, from data import, cleaning, exploration, model training, to online real-time inference, supporting the complete AI data processing process.
Independent Deployment Monitoring Tool
It is equipped with an independent deployment monitoring tool, which can be used to monitor system resources, data status, and cluster information in real-time. It can be integrated with multiple alert methods to assist with daily operations and maintenance work.
eZoo Graph ADVANTAGE
Triple Data-Import Performance
In a data set with a scale of 2 million nodes and 70 million edges, eZoo completed the import in just 20 seconds, with an efficiency of over 3 times that of a leading company.
Triple Algorithm Efficiency
At a scale of 100 million nodes, 5 billion edges, and 130 GB of data, eZoo completes the PageRank algorithm in just 1.5 minutes, which is more than three times faster than other mainstream graph databases.
Triple Data-Import Performance
In a data set with a scale of 2 million nodes and 70 million edges, eZoo completed the import in just 20 seconds, with an efficiency of over 3 times that of a leading company.
Triple Algorithm Efficiency
At a scale of 100 million nodes, 5 billion edges, and 130 GB of data, eZoo completes the PageRank algorithm in just 1.5 minutes, which is more than three times faster than other mainstream graph databases.
Save Costs By Consuming 1% of Time
In the fifth-degree directed path test, eZoo's graph database query speed is 500 times faster than that of leading company A; In the bidirectional path test with a timeout setting of 300 seconds, eZoo's graph database query speed is more than 100 times faster than that of mainstream leading company A.
Over 100 Times Faster on Neighborhood Query
In the forward neighbor query test, eZoo's speed is 100-1000 times faster than that of mainstream leading companies. In the bidirectional neighbor query test for returning the number of neighbors, eZoo's query speed is more than 500 times faster than that of leading companies, taking only 0.727ms, while the other two companies take 375.5ms and 8.035ms, respectively.
Save Costs By Consuming 1% of Time
In the fifth-degree directed path test, eZoo's graph database query speed is 500 times faster than that of leading company A; In the bidirectional path test with a timeout setting of 300 seconds, eZoo's graph database query speed is more than 100 times faster than that of mainstream leading company A.
Over 100 Times Faster on Neighborhood Query
In the forward neighbor query test, eZoo's speed is 100-1000 times faster than that of mainstream leading companies. In the bidirectional neighbor query test for returning the number of neighbors, eZoo's query speed is more than 500 times faster than that of leading companies, taking only 0.727ms, while the other two companies take 375.5ms and 8.035ms, respectively.
eZoo Graph CAPABILITIES
Multi-threaded Parallel Computing
Based on a high-performance parallel computing framework, utilizing high compression and contiguous memory to achieve non-blocking and weak preemption parallel computing.
Distributed Computing
Based on the RAFT protocol to enhance the high availability and query performance of the database, supporting one master and multiple replicas, automatic disaster recovery, and read-write separation.
Graph Computing
Implemented graph mining algorithms such as shortest path, PageRank, Louvain, LPA, K-core, WCC, SCC based on a high-performance graph computing platform.
Transaction Enabled
Support for transaction isolation levels of read_uncommitted and serializable.
Data Storage
Data storage is achieved through two structures: LSM-tree and in-memory mirroring.
Supported Language
Compatible with OpenCypher query language.
UDF and SDK
High-performance C++ UDF (User Defined Functions), Java, Python, and Restful API SDK.
Deployment
We provide images for the graph database and visual platform based on CentOS, as well as deployment configuration files for the graph platform based on docker-compose/k8s.
Data Preloading
Key data information will be preloaded into memory to provide the basic data for high-performance parallel computing.
Memory Compression
Memory compression is achieved through custom variable-length storage structures, intelligent mapping, encoding algorithms, and efficient template design.
eZoo Graph
The World's Leading Graph Intelligence Platform
Copyright @2022 – 2023 EZOO GRAPH AI INC. All Rights Reserved.