Hologres Shared Cluster (Lakehouse Acceleration Edition) is a serverless offering that accelerates queries on MaxCompute and OSS data at low cost. This topic covers its background, scenarios, features, and a comparison with exclusive instances.
Background information
Hologres Shared Cluster (Lakehouse Acceleration Edition) is an upgraded version of Hologres Shared Cluster (BI-based MaxCompute Acceleration Edition). In addition to BI-based MaxCompute acceleration, it supports OSS data lake acceleration. It adopts a cloud-native architecture that separates storage from computing. Each instance has exclusive frontend nodes with 4 cores and 16 GB of memory for SQL authentication, parsing, and optimization, while sharing a computing resource pool to deliver high computing power at low cost. This enables cost-effective, serverless interactive analysis across data lakehouses.
Scenarios
Hologres Shared Cluster (Lakehouse Acceleration Edition) is applicable to the following scenarios:
-
Data stored in OSS or MaxCompute is queried infrequently and requires low latency.
-
Complex queries on large volumes of foreign table data exceed the capacity of low-specification exclusive Hologres instances.
-
Data lake and data warehouse queries need acceleration, with BI tools such as Tableau, Quick BI, and FineBI connected for analytics.
Limits
Hologres Shared Cluster (Lakehouse Acceleration Edition) is supported in the following regions: China (Beijing), China (Shanghai), China (Hangzhou), China (Shenzhen), and Singapore.
If a Hologres Shared Cluster instance does not receive any query request within 90 consecutive days, it is suspended at 12:00 on the 90th day to optimize resource allocation. After the suspension, the metadata and endpoint of the instance remain unchanged, but the computing service is paused and no computing fees are incurred. To resume the instance, go to the Instances page.
Features
Hologres Shared Cluster (Lakehouse Acceleration Edition) instances share computing resources but offer the same rich feature set as exclusive Hologres instances. Each instance provides the following features:
-
Built-in query engine:
-
You can create foreign tables to accelerate queries on data stored in MaxCompute and OSS. The data remains in the external system.
-
You cannot create internal tables or import data into Hologres Shared Cluster (Lakehouse Acceleration Edition) instances. To use internal tables, purchase an exclusive Hologres instance.
-
-
Standard SQL statements: Compatible with PostgreSQL 11, Hologres Shared Cluster (Lakehouse Acceleration Edition) supports standard PostgreSQL statements for data queries.
-
Permission management: Permissions are managed per instance, the same as for exclusive Hologres instances. You can add or remove users and grant different access permissions based on your business requirements.
Hologres Shared Cluster (Lakehouse Acceleration Edition) instances support all features of exclusive Hologres instances except internal table and data import features.
Comparison of service types
The following table compares exclusive Hologres instances and Hologres Shared Cluster (Lakehouse Acceleration Edition) instances.
|
Item |
Exclusive Hologres instance |
Hologres Shared Cluster (Lakehouse Acceleration Edition) instance |
|
|
Feature |
Category |
PostgreSQL 11 is supported, with more operators available. |
PostgreSQL 11 is supported. |
|
Parameter optimization |
Parameter optimization |
Supported. You can modify database-level or user-level parameters based on your business requirements. |
Supported. You can modify database-level or user-level parameters based on your business requirements. |
|
Number of SQL connections or queries per second (QPS) |
Varies by instance type. Run the |
The default number of SQL connections is 100. |
|
|
Data query based on foreign tables |
Meta module loading |
Queries can be accelerated to respond within milliseconds. |
Queries can be accelerated to respond within sub-seconds. |
|
Data type |
The basic data types, DECIMAL type in MaxCompute V2.0, DATETIME type, and ARRAY type are supported. |
The basic data types, DECIMAL type in MaxCompute V2.0, DATETIME type, and ARRAY type are supported. |
|
|
Engine optimization |
N/A. |
N/A. |
|
|
Data query based on internal tables |
Supported indexes |
Bitmap indexes and clustering keys are supported. |
None. |
|
Specified distribution keys |
Distribution keys can be specified for Local Join or Local Group By operations. |
None. |
|
|
Resources |
Purchased computing and storage resources are exclusive to you. |
Computing resources are shared; storage resources are exclusive to you. |
|