INFOWORLD | AUGUST 5, 2014
Today's one-size-fits-all database processing will give way to a cheaper, but more complex heterogeneous approach. This idea has gotten new attention because it is the core idea behind MapReduce, the parallel processing model used by Hadoop in big data analytics. These types of distributed workloads have been used for years, typically with a homogeneous server cluster, meaning it works across lots of the same servers. That homogeneity restricts you to one server cluster or one cloud -- thus, one resource type and cost. But not any more.
No comments:
Post a Comment