Google Cloud and intel collaborated to develop a new chip category that Intel thinks would become a key selling in the burgeoning cloud computing business.
The new chip, known as Mount Evans, will be marketed to companies other than Google and mirrors how cloud services work. They create massive data centers full of powerful physical computers and sell virtual chunks of those machines to other enterprises, who earn a greater return on investment than building the equipment themselves. Establishing virtual computers and sending client data to the correct location are essentially overhead expenditures for cloud providers.
The Mount Evans chip, dubbed an “infrastructure processing unit” (IPU) by Google and Intel, isolates those operations from the core computational processes and speeds them up. This also helps to protect certain functions from hackers and gives flexibility to the data center. This is critical from a strategic standpoint, in our opinion. “It’s a critical area for us or the data centre,” said Nick McKeown, senior vice president of Intel’s network and edge division, to Reuters. Intel wasn’t the only company that manufactures infrastructure chips. Nvidia and Marvell Technology both have offers that are comparable yet slightly different.
However, Intel and Google are collaborating on free software tools to make Intel’s version of the chip a larger industry-standard used outside of Google’s data centers.
According to Amin Vahdat, a Google fellow and vice president of engineering, Google hopes to kickstart a technological trend that allows all data center operators to be more flexible in splitting up their physical computer servers into virtual ones to meet whatever computing work is at hand.
“The fundamental question of what constitutes a server will extend beyond what is contained within the sheet metal. The IPU will play a key part in this “According to Vahdat.
