O'Reilly logo
  • Shawqi Mohammed thinks this is interesting:

fixed compute and storage coupled together can lead to low utilization and a lack of flexibility. The problem with this is that not all Hadoop jobs will use the same level of compute and storage resources. If there isn’t enough compute power or there is a lack of storage resources, the solution is often to add more nodes. Separating compute and storage resources can lead to better utilization of hardware, decreasing the tendency to add more nodes when there are bottlenecks in just one place. Creating multiple virtual machines that are right sized on the same ESXi server can also provide performance advantages.


Cover of Virtualizing Hadoop: How to Install, Deploy, and Optimize Hadoop in a Virtualized Architecture