Google Cloud has launched its first line of Arm-based VMs, making it the third endeavour after the AWS-Graviton and Azure-Arm deals.
The latest addition, Tau T2A under the Tau VM family, will use Ampere-based Altra Arm processors with single-threaded performance – a necessity for Oracle, Tencent, Azure and Equinix Metal processors.
Tau T2A VMs offers numerous predefined SKUs, networking bandwidth of 32 GBPS, 48 vCPUs per VM, 4GB of memory per vCPU, storage options, and improved performance of up to 56% over Azure and AWS.
Google Cloud comments: “Arm-based chips are already ubiquitous in mobile devices, and have proven themselves for supercomputing workloads. At Google Cloud, we’re also excited about using Arm chips for the next generation of scale-out, cloud-native workloads.”
According to Google Cloud, it created Tau to deliver maximum performance in running cloud-native, scale-out environments, including but not limited to containerised workloads, Java-based applications, media transcoding, data-logging processing and web servers. The new Tau version comes with a trial program for developers to run and execute workloads while receiving the necessary training to understand the VM.
Nonetheless, the downside includes no memory extension support, nested virtualisation, sole tenancy and reduced customisation. With a Tau and Arm crossover, Google Cloud might face difficulties handling legacy workloads.
Arm supports openness and Tau being an x86 architecture, offers limited compatibilities with devices using a non-x86. On the other hand, Arm will deploy Neoverse N1-based T2A to power Tau alongside Google Kubernetes for containerised workloads.
- Arm Shows Financial Muscle with Record Revenues – read the news here
The new deal is set to build chips from a price standpoint. The basic pricing plan for a 32-score Tau T2A VM costs around $1.232 (£1.038) per hour, a definite plus compared to key competitors. The newly designed VM is compatible with such operating systems as Ubuntu, CentOS, RHEL, Rocky Linux and Google’s own Container-Optimized OS.
Besides VMs, Google is also looking to partner with Arm over Batch and Dataflow Stream. With Batch, Google is planning to automate human resources and job listings through analysing job queues, autoscaling resources, executing subtasks and running jobs, while debugging common errors without building a full-fledged engineering team.
The partnership could be seen as a huge blow to AMD and Intel and their duopoly over the chip manufacturing vertical.
SoftBank-owned Arm closed last year at a record $2.7 billion (£2.2 billion) revenue and 29.2 billion chips.
Last month, Arm chief executive Rene Haas told the Financial Times that the UK chip designer would seek to step up its push beyond mobile phones and deeper into cars, data centres and hardware underpinning the metaverse. The FT notes that SoftBank is inching closer to pursuing a dual listing of Arm in both New York and London after months of intense lobbying by the UK government.
Prior to that, Arm announced in March it will cut between 12-15% of its workforce due to the collapse of NVIDIA’s $40 billion (£29 billion) takeover.