×
Community Blog Alibaba Cloud Announced Major Upgrades of GanosBase and Released the First Cloud Twin Spatio-Temporal Database

Alibaba Cloud Announced Major Upgrades of GanosBase and Released the First Cloud Twin Spatio-Temporal Database

This article discusses GanosBase V4.0 and the benefits of the first cloud twin spatio-temporal database.

1

Introduction

GanosBase is a new generation location intelligence engine developed jointly by Alibaba DAMO Academy's Database and Storage Lab (led by Feifei Li) and Alibaba Cloud. The first cloud twin spatio-temporal database was launched in the major upgrade V4.0.

Multi-dimensional spatial data represented by the Building Information Model (BIM), Geographic Information System (GIS), and spatio-temporal moving objects has exploded and become an important foundation for new infrastructure and digital framework with the rapid development of earth observation technology, Internet of Things (IoT), and digital twin technology. From digital twin factories, digital twin cities, and digital twin earth, multi-dimensional spatial data is increasing constantly, which brings challenges to data storage, intelligent processing, analysis, and computing.

Cloud computing promotes the rapid evolution of databases towards cloud-native and provides enormous computing power for the combination of database technology and digital twin technology. Databases usher in a new era from cloud-native to cloud twin as the core system of data processing. The cloud twin spatio-temporal database released this time in GanosBase V4.0 improves the storage and processing capabilities of high-dimensional and dynamic spatial data significantly. The underlying layer supports different database products, such as ApsaraDB RDS for PostgreSQL, PolarDB for PostgreSQL, AnalyticDB for PostgreSQL, ApsaraDB for Lindorm, and Data Lake Analytics (DLA).

2
GanosBase is compatible with different database products.

1. What Is a Cloud Twin Spatio-Temporal Database?

Cloud-native is an approach to design, systems, and applications built on the cloud, taking full advantage of the cloud platform with scalability and distributed architecture. Digital twins are simply a mirror digital description of the physical world. Specifically, the physical world is moved into a computer. On the other hand, cloud twins combine digital twins with the cloud and take the cloud as the foundation of digital twin systems to design the systems based on the cloud-native architecture. Cloud twins are a more advanced form after the combination of digital twin systems and cloud-native technologies. It is an inevitable trend led by the digital technology of all things.

GanosBase is the first cloud twin spatio-temporal database built based on cloud-native databases and digital twin engine technology. The system contains three design philosophies: a strong twin kernel, integrated big storage, and fast big computing.

The strong twin kernel implants the digital twin engine into the kernel of the cloud-native database system to integrate new complex scenario data and high-dimensional multimodal computing.

Integrated big storage is based on a unified spatio-temporal location framework and uses a multi-dimensional database engine newly built from the bottom up to achieve centralized one-stop management of BIM + GIS + IoT superfusion data.

Fast big computing is based on multi-dimensional entity coding and efficient compression algorithms, combined with an elastic multi-level parallel computing framework. This makes multi-dimensional data accessible for humans and machines and improves the efficiency of large-scale 3D and 4D analysis and computing a hundred times higher than traditional software.

2. Eight Professional Engines Enable Multimodal Superfusion

A single model can no longer support the new application and scenarios of digital twins. GanosBase combines the cloud-native database system to develop a multi-dimensional and multimodal data processing engine from zero and the kernel. As of V4.0, the system has provided eight core engines for geometry, raster, trajectory, 3D, point cloud, path, mesh, and fast display. GanosBase enables the database to support efficient integrated storage, query, analysis, and computing of more than ten categories of multi-dimensional multimodal data natively through continuous upgrading and improvement.

3
Eight Core Engines of GanosBase

3. Interpretation of Five Key Features

Redefining 3D Spatial Computing

The digital and accurate mapping of various physical entities, such as buildings, road facilities, machinery, and equipment cannot be separated from 3D spatial modeling. GanosBase V4.0 has introduced a new 3D engine and has incorporated the storage, indexing, query, and analysis processing capabilities of multi-dimensional spatial data, such as BIM and real-life 3D, into the cloud-native database. The analysis and computing efficiency in complex scenarios is increased a hundredfold. The system adopts the completely self-developed meshflex "flexible mesh" geometry technology, which includes flexible 3D geometry and scenario expression methods. The spatial geometry organization adopts multi-level structure expression, and the basic geometric elements can be nested and combined to realize complex geometric models. Built-in 3D spatial indexes in the database are supported, and tens of millions of 3D entities can be indexed in a few milliseconds. The system supports 3D spatial measurement, spatial analysis, 2D-3D-4D integrated analysis, and computing. The analysis efficiency of tens of millions of polygon elements is more than 100 times faster than traditional software.

4
Meshflex Flexible Mesh Geometric Modeling and Efficient 3D Computing

Advanced Mobile Object Processing

Moving objects, such as people, cars, ships, and aircraft, are similar to blood flowing in the physical world. They are essential in building a digital twin system. GanosBase is the first cloud-native 4D (3D space + time) moving object database, which can store, index, and analyze multimodal data, such as spatio-temporal information, thematic attributes, and behavioral events of moving objects. V4.0 provides two major feature upgrades:

  • Tiered Storage of Cold and Hot Moving Objects: The system transparently stores specified moving object data in a table from offline to external storage media, such as Object Storage Service (OSS), and allows transparent SQL query and access. This feature reduces storage costs significantly while keeping all data online. Taking AIS ship trajectory data as an example, more than 90% of storage space can be saved.
  • Trajectory Segmentation Index: GanosBase adopts an internal trajectory segmentation index policy for moving object trajectories with a large space-time span. The test shows that the efficiency of index filtering has improved by more than five times, improving the performance of 4D trajectory query and analysis.

5
GanosBase mobile object database supports 4D spatio-temporal analysis for drones.

Multi-Level Spatial Parallel Query Acceleration

Analysis, control, and simulation are the key elements of digital twin systems. It is difficult for traditional systems to achieve real-time and fast topological analysis and calculation for large buildings or large-scale 3D entity objects. Based on the cloud-native database PolarDB, GanosBase has established a multi-level spatial parallel query processing framework, which supports intra-parallel (cross-node parallel), inter-parallel (inter-node table-level parallel), and obj-parallel (object-level parallel). The processing efficiency of large-scale multi-dimensional spatial data analysis is improved by at least one order of magnitude.

6
PolarDB + GanosBase Multi-Level Spatial Parallel Query Processing Framework

Based on the test data of a national-level new area digital planning system, it is shown that in the complex scenario where ten million-level elements of large-scale buildings are analyzed and computed, the performance of multi-level spatial parallel computing is more than a hundred times higher than traditional plans. The system can be applied to smart city, CIM, and land scenarios where the data volume is massive and the objects are high-dimensional and complex to accelerate spatio-temporal computing.

7
BIM large building complex analysis and computing is more than 100 times more efficient than traditional software.

Distributed Aerospace Computing

Taking images of Earth through satellites is one of the most direct and effective ways to obtain data for future digital twin scenarios. However, due to different sensors and different shooting times of satellites, images in different regions often overlap with each other, and the colors are discontinuous, which affects the real effect and machine-based intelligent interpretation. GanosBase integrates PolarDB's aerospace index capability and DLA Spark's aerospace computing capability. In V4.0, GanosBase provides automatic, large-scale, and distributed spatio-temporal editing of remote sensing images and uniform color algorithm services, supports automatic extraction of mosaic lines and contribution areas, supports a variety of industry-leading automatic uniform color schemes, and can generate a map based on globally uniform colors automatically. GanosBase integrates the AI Earth product of Alibaba DAMO Academy to form a DB for AI solution, which has been widely used in many fields, such as natural resources, environmental monitoring, and agricultural production.

8
GanosBase aerospace raster engine comprehensively supports AI Earth.

PB-Level Trajectory Big Data Processing

Combined with the multimodal database ApsaraDB for Lindorm, GanosBase provides 100 TB to PB-level trajectory big data storage and processing capabilities, mainly serving new fields and scenarios, such as the Internet of Vehicles, shared mobility services, navigation and aviation, and public security. The following core capabilities are enhanced in this upgrade:

  • Flexible Trajectory Storage Model: The system supports three storage models, including the point model, whole trajectory model, and segmentation trajectory model. This is suitable for different query and analysis scenarios.
  • Low-Cost Trajectory Storage: Combined with native secondary indexes of ApsaraDB for Lindorm, the system ensures query efficiency and reduces storage costs to 50% of the full index storage model. This is suitable for storing large amounts of trajectory data.
  • Efficient Spatio-Temporal Query: The system supports four types of common trajectory queries, including ID + time query, attribute query, spatial range query, and spatio-temporal range query. You can build secondary indexes to support multi-dimensional queries, and the response time is at the 10-millisecond level.

On this basis, the system integrates with ApsaraDB for Lindorm to support the integrated management and processing of multimodal and spatio-temporal data, such as key-value data, wide tables, time-series data, search data, and files. This provides a more comprehensive foundation for the construction of the digital twin system.

9
GanosBase supports integration with ApsaraDB for Lindorm.

The release of the cloud twin spatio-temporal database GanosBase will provide a new solution for analysis in large-scale multi-dimensional complex scenarios and the processing of multimodal fusion data. Jiong Xie, Head of GanosBase Technology Research and Development, points out that to deliver the processing capability of spatial 3D data in real-world scenarios to the cloud-native database, the core is to address the issue of data storage of management during data query and computing. Data computing here refers to 3D and 4D multi-dimensional analysis and processing capabilities. In the process, multidimensional data is put into practical application. The process is similar to movie production. Data visualization is like special effects of scenes, and analysis and processing capabilities are like the story told in the movie. If the movie only presents flamboyant scenes but lacks profound meaning behind the scenes, it cannot be a good movie. GanosBase V4.0 is unveiled. The combination of cloud-native technology and cloud twin technology will contribute to a good movie for digital twin applications.

0 0 0
Share on

ApsaraDB

152 posts | 12 followers

You may also like

Comments

ApsaraDB

152 posts | 12 followers

Related Products