E-MapReduce (EMR) provides a cloud-hosted service for ClickHouse, an open source Online Analytical Processing (OLAP) engine. The EMR ClickHouse service is fully compatible with the open source version and provides cloud-native features such as rapid cluster deployment, cluster management, scale-out, scale-in, and monitoring and alerting. It also optimizes read and write performance and enhances integration with other EMR components.
Features
|
Feature |
Description |
|
Columnar storage |
Columnar storage delivers better query performance than row-based storage and achieves a higher data compression ratio, which saves storage space. |
|
MPP architecture |
Each node accesses only its local memory and storage, and nodes interact and process data in parallel. This design delivers high query performance and scales easily. Vectorized engine: Data is stored by column and processed in vectors (parts of a column), enabling more efficient CPU utilization. |
|
SQL support |
ClickHouse supports a declarative, SQL-based query language similar to the ANSI SQL standard, including GROUP BY, ORDER BY, FROM, JOIN, and IN queries, along with non-correlated subqueries. |
|
Real-time data updates |
ClickHouse lets you define primary keys in tables. Data is stored in an ordered, incremental way in the MergeTree, enabling fast range lookups on the primary key. ClickHouse supports near-real-time data insertion, metric aggregation, and index creation. |
|
Index support |
ClickHouse sorts data by the primary key, which allows it to locate specific values or data ranges in tens of milliseconds. |
Typical Scenarios
|
Scenario |
Description |
|
User behavior analysis |
Create large wide tables for behavior analysis systems, with each table containing over a thousand columns. Fewer JOIN operations are needed, and the design supports path analysis, funnel analysis, and path conversion analysis. |
|
Traffic and monitoring |
Use stream compute engines such as Flink or Spark Streaming to clean and process system and application monitoring metrics, write the data to ClickHouse in real time, and visualize the results with Grafana. |
|
User profile |
Transform various user features to create one or more feature tables for all users. These tables support flexible user profile analysis for business needs such as advertising and user segmentation. |
|
Real-time BI reports |
Create flexible, real-time BI reports to meet business needs. Most queries return results in seconds. Examples include order analysis, marketing effectiveness analysis, and sales promotion analysis. |
Unsupported scenarios:
-
It does not provide full support for transactions.
-
It cannot modify or delete existing data at a high frequency and with low latency.
-
It is suitable only for batch deletion or modification of data.