Google Cloud will become the latest company to run its data centre workloads on chips based on IP from British designer Arm, it has announced. It is the latest sign that Arm designs are beginning to gain traction in the data centre market, which is dominated by x86-based processors from Intel and AMD.
A new family of virtual machines (VMs), Tau T2A, is Google Cloud’s first based on Arm architecture. The company announced the general availability of the machines at its Infrastructure Spotlight event on Wednesday.
The news means the big three public cloud providers – Google Cloud and its rivals Amazon’s AWS and Microsoft Azure – all run Arm-based chips in their data centres.
What are the Google Cloud T2A virtual machines for?
The latest iteration of the Tau VM range it launched last year, Google says the Tau T2A VMs will offer a wide range of network-attached storage options, and can be used for scale-out workloads including web servers, containerised microservices, data-logging processing, media transcoding, and large-scale Java applications.
“Developers now have the option of choosing the optimal architecture to test, develop and run their workloads,” said Sachin Gupta, vice president and general manager for infrastructure at Google Cloud.
Tau T2A VMs run on Altra processors from US chip company Ampere. Founded in 2018 by a group of former Intel executives, the company designs server chips built on Arm architecture.
Google’s Arm server chips and changing data centre chip market
Google will still run many of its data centres on chips built on the x86 architecture pioneered by Intel, but its decision to offer Arm-based VMs reflects the growing interest in using the British company’s designs in data centres too.
In April, Microsoft announced it was offering VMs powered by Ampere Altra CPUs to customers of its Azure cloud service. Ampere also counts Oracle among its customers, as well as Chinese tech giants Tencent, and Baidu. Last month HPE announced it would become the first vendor to start selling servers running Ampere chips to its customers.
Other companies are using Arm designs to produce their own server chips, with Amazon and Tencent both working on custom silicon to power the data centres needed for their cloud services. Amazon’s cloud platform, AWS, launched services based on its latest processor design, Graviton3, in May, and claims it offers a 25% performance increase on previous versions.
Despite this shift, the data centre server market remains dominated by designs built on the x86 architecture, with the vast majority of server chips built by Intel or its US rival AMD. The data centre server tracker report from analyst company Omdia shows that in Q3 2021, around 5% of servers, some 170,000 units, shipped with an Arm chip.
Keen to preserve its market share, Intel has been fighting back, with a new range of chips for servers announced in May, including graphical processing units for AI workloads.
Tech Monitor is hosting a roundtable in association with Intel vPro on how to integrate security into operations. For more information, visit NSMG.live.