Say what you want about Dell, but don’t say it doesn’t understand the hardware architectures future applications will want to run on. On Wednesday, the server giant announced it is donating a concept version of an ARM-processor-based server design to the Apache Software Foundation so contributors to the foundation’s various projects — including big data projects such as Hadoop and Cassandra — can optimize them to run on the low-power gear.
Big data is actually a ripe field for ARM-server manufacturers to pick, as scaling out to handle large volumes of data can also mean using up a lot of energy. The problem can get so bad at scale, in fact, that web analytics company Quantcast actually developed its own distributed file system for Hadoop to help cut down on the power bills. The theory behind Dell’s new server concept — which is called “Zinc” and uses Calxeda’s EnergyCore architecture — is that parallel processing jobs that need a lot of nodes, but not necessarily a lot of brawny nodes, can do the same work while consuming far less power.
Calxeda’s Barry Evans explained it thusly last year, when commenting on competitor SeaMicro’s deployment of an Intel Atom-based Hadoop cluster at eHarmony:
“Big data is a great fit for us and ARM servers for three key reasons. First, it is an inherently scale-out application, requiring a lot of efficient processors. Second, it is a fast-growing market place without a lot of requirements for legacy baggage. Third, the application software is widely available to run on ARM today.”
Zinc isn’t Dell’s first foray into the world of low-power servers, either. In May, it announced another ARM-based design called “Copper,” and it has been rolling out a collection of low-power Intel Xeon servers for the past couple years that are modeled on the hardware it sells to its webscale Data Center Solutions customers. One could say this all evidence that Dell understands it’s part of a changing server market where fewer companies, such as Facebook, will buy a greater percentage of servers, and where vendors had better meet their efficiency demands or get left out in the cold.