To meet the demand for storage in big data scenarios, Alibaba Cloud has launched the D1 series on the cloud.

Instead of using cloud disks in its data storage, the D1 series uses ephemeral disks, which solves the problem of high costs caused by keeping multiple copies of redundant data in cloud disks. Data also no longer needs to be transferred over the network, which improves disk throughput. Furthermore, with the D1 series, you can also take advantage of Hadoop's proximity computing.

Compared with cloud disks, the series greatly enhances storage performance while reducing prices. Indeed, the cost is almost the same as offline physical hosts.

Despite their advantages, however, ephemeral disks still cannot ensure data reliability, and upper-layer software is required to guarantee it. If a disk or node fails, operations and maintenance must be performed manually. Cloud disks, meanwhile, guarantee data reliability automatically, meaning that you do not need to worry about disk damage. Alibaba Cloud's default multi-disk backup policy is also helpful in this regard.

E-MapReduce + D1 solution

A complete set of automated O&M solutions, such as the D1 series, is now available for ephemeral disks in E-MapReduce. This allows Alibaba Cloud users to use ephemeral disks conveniently and reliably, without having to worry about the entire O&M process. Data reliability and service availability are guaranteed.

The main advantages are as follows:

  • Highly reliable distribution of required nodes
  • Ephemeral disk and node fault monitoring
  • Automatic determination of data migration opportunities
  • Automatic failed node migration and data balancing
  • Automatic HDFS data detection
  • Network topology optimization
With the automated O&M of the entire back-end management and control system, E-MapReduce helps you make better use of ephemeral disks and develop a cost-effective big data system.
Note If you want to set up a Hadoop cluster using the D1 series, submit a ticket. We will then be able to assist you in your operations.