Help Center/ Data Lake Insight/ FAQs/ Product Consultation/ Usage/ What Are the Differences Between DLI Flink and MRS Flink?
Updated on 2023-05-19 GMT+08:00
View PDF

What Are the Differences Between DLI Flink and MRS Flink?

DLI Flink suits the cloud-native infrastructure. DLI Flink optimizes multiple core functions of the kernel engine. DLI provides an enterprise-level one-stop development platform with built-in development and O&M functions, freeing you from the trouble of O&M of user-built clusters. In addition to open-source connectors, DLI Flink also supports MySQL, GaussDB, MRS HBase, DMS, GaussDB(DWS) and OBS on the cloud, which allows you use DLI Flink right after service subscription. More importantly, DLI Flink jobs can automatically adapt to service requirements. Resources can be automatically and flexibly scaled to ensure service stability.

Table 1 compares the functions of DLI Flink and MRS Flink.

Table 1 Function comparison

Category

Features

DLI Flink

MRS Flink

Featured Capabilities

Deployment

Full hosting

Semi-hosting (cluster O&M required)

Elastic scaling
  • Container-based cluster deployment
  • Users can perform elastic scaling based on the service load and dynamically adjust the resources based on the job load.
  • Resources used by jobs can be dynamically adjusted based on job priorities.

Only YARN clusters are supported.

Upstream and downstream data connection
  • Open-source connectors and out-of-the-box connectors for data sources including databases (RDS and GaussDB), message queues (DMS), data warehouses (GuassDB DWS), and object storage (OBS).
  • Higher usability and stability (in comparison with open-source connectors)

Open-source connectors only

Development and O&M

Monitoring and alarms
  • Interconnection with Cloud Eye, a monitoring platform on Huawei Cloud, and SMN system on the cloud. Users can be notified of any events through emails, SMS messages, calls, and third-party office tools (webhook)
  • Interconnection with the unified monitoring and alarm system (Prometheus) of companies
  • Flink job rate, input and output data volume, operator backpressure, operator delay, CPU usage, and memory usage

Flink UI alarms only

Multi-versioning

Different Flink versions can be used for jobs.

Jobs of different Flink versions are not supported by a Flink cluster.

Usability

Out-of-the-box, serverless architecture, and cross-AZ disaster recovery

  • Users only need to write SQL statements and do not need to compile them. You only need to pay attention to the service code.
  • You can use SQL statements to connect your jobs to various data sources, such as RDS, GaussDB(DWS), Kafka, and Elasticsearch.
  • Users do not need to log in to the maintenance cluster. They can submit jobs in few clicks on the console.
  • Checkpoints for Flink SQL jobs
  • Flink job logs can be dumped and stored for job analysis.

Technical capabilities are required for coding, cluster setup, configuration, and O&M.

  • Users need to write and compile code.
  • Users need to log in to the cluster and run commands to submit the configuration. They need to maintain the cluster.
  • Users need to code checkpoints to enable the function.

Job templates

Preset general Flink SQL templates for quick start

N/A

Enterprise security

Configuring access control

Permission control streamlined with Huawei Cloud IAM, and role-based access control

N/A

Space isolation

Tenant-level and project-level isolation resources and code for efficient team works

N/A
Parent topic: Usage