Business Assessment

Prev Next

search by keyword

Recent Pages

Documentation

Download PDF

Last updated: 2021-07-08 10:43:44

Download PDF

Selecting a Cluster Type

Elastic MapReduce (EMR) provides five types of clusters for you to choose from based on your business needs.

Hadoop cluster: based on open-source Hadoop and the components that form a Hadoop ecosystem, it provides big data solutions for massive data storage, offline/real-time data analysis, streaming data compute, and machine learning.
Druid cluster: Druid is a high-performance real-time analytics database. It supports big data queries in milliseconds and multiple data ingestion methods. It is suitable for real-time big data query scenarios.
ClickHouse cluster: ClickHouse is a column-oriented OLAP database management system. It is suitable for data warehouse analysis scenarios such as real-time wide table analysis, real-time BI report analysis, and user behavior analysis.
Doris cluster: Doris is an MPP analytical database product that supports sub-second queries on PB-scale, structured data. It is compatible with MySQL protocol and uses the standard SQL syntax. It is suitable for historical report analysis, real-time data analysis, interactive data analysis, etc.
Kafka cluster: Kafka is a distributed, partitioned, multi-replica, and multi-subscriber message processing system based on ZooKeeper coordination. It is suitable for asynchronous processing, message communication, and streaming data receiving and distribution.

Selecting a Billing Mode

Billing mode for EMR clusters:

Pay-as-you-go: all nodes in a cluster are charged on a pay-as-you-go basis. This is suitable for clusters that exist for a short time or periodically.

For information on node types, see Node Type Description.

Selecting a Model and Specification

EMR offers a wide variety of CVM models, including EMR Standard, EMR Compute, EMR High IO, EMR MEM-optimized, and EMR Big Data. If you need the CPM model, please submit a ticket to us.

You can choose a model based on your business needs and budget.

If you require low latency for offline compute, we recommend you select a model with local disks or the Big Data model.
If you need to use the real-time database HBase, we recommend you select the EMR High IO model with local SSD disks for optimal performance.

Node specification recommendations

Cluster Type	Node Type	Recommended Specification
Hadoop	Master	You are advised to select an instance specification with a large memory size (at least 8 GB) and use cloud disks for high stability.
	Core	If most of your data is stored on COS, core nodes will function in a way similar to task nodes and should have a capacity of at least 500 GB. Core nodes cannot be elastically scaled. If your architecture does not use COS, core nodes are responsible for processing cluster compute and storage tasks, and three-replica backup is enabled by default. When estimating the data disk capacity, you need to consider the capacity for storing three replicas. In this case, the Big Data model is recommended.
	Task	If your architecture does not use COS, task nodes are not required. If most of your data is stored on COS, task nodes can be used as elastic compute resources and deployed as needed. If the billing mode of your cluster is monthly subscription, task nodes need to be pay-as-you-go.
	Common	Mainly used as ZooKeeper nodes. You are advised to select the specification of 2 cores, 4 GB memory, and 100 GB cloud disk capacity to meet the requirements.
	Router	Mainly used to relieve the load of master nodes and as a task submitter. Therefore, you are advised to select a model with a large memory size, preferably not lower than the specification of master nodes.
ClickHouse	Core	You are advised to select a model with high CPU and a large memory size. Because data may be lost if a local disk is corrupted, cloud disks are recommended.
ClickHouse	Common	The CPU and memory configuration should be at least 4 cores and 16 GB.
Kafka	Core	You are advised to select a model with high CPU and a large memory size. Because data may be lost if a local disk is corrupted, cloud disks are recommended.
Kafka	Common	The CPU and memory configuration should be at least 4 cores and 16 GB.
Doris	Master	You are advised to select an instance specification with a large memory size (at least 8 GB) and store all the metadata of master nodes in the memory.
	Core	You are advised to select an instance specification with a large memory size (at least 8 GB) and use cloud SSD for better IO performance and stability.
	Router	The frontend module is deployed here for high read/write availability. Therefore, you are advised to select a model with a large memory size, preferably not less than that of master nodes.
Druid	Master	You are advised to select an instance specification with a large memory size (at least 16 GB) and use SSD for better IO performance.
	Core	You are advised to select an instance specification with a large memory size (at least 8 GB) and use cloud SSD for better IO performance and stability.
	Task	If your architecture does not use COS, task nodes are not required. If most of your data is stored on COS, task nodes can be used as elastic compute resources and deployed as needed.
	Common	Mainly used as ZooKeeper nodes. You are advised to select the specification of 2 cores, 4 GB memory, and 100 GB cloud disk capacity to meet the requirements.
	Router	Mainly used to relieve the load of master nodes and as a task submitter. Therefore, you are advised to select a model with a large memory size, preferably not lower than the specification of master nodes.

Network and Security

To ensure the network security, the EMR cluster is placed in a VPC, and a security group policy is added to the VPC. In addition, to ensure easy access to the WebUI of Hadoop, a public IP is enabled for one of the master nodes and the node is billed by traffic. A public IP is not enabled for router nodes by default. However, you can bind a router node to an EIP on the CVM console to enable a public IP for it.

Note：

A public IP is enabled for master nodes when a cluster is created. You can disable it as needed.

Enabling a public IP for master nodes is mainly for SSH login and component WebUI access.
Master nodes with a public IP enabled are billed by traffic with a bandwidth of up to 5 Mbps. You can adjust the network on the console after creating a cluster.

Recent Pages