Google Cloud is significantly enhancing its Arm-based computing capabilities with the introduction of its new Axion N4A instances. Marked as a game-changer, these instances aim to reduce the dependence on traditional x86-based instances and provide cost-effective alternatives for enterprises.
The N4A series targets a variety of general-purpose workloads, including low-to-medium traffic web and application servers, containerized microservices, virtual desktops, and small-to-medium databases. Available in preview, these instances can be configured with up to 64 virtual CPUs, support 512GB of DDR5 memory, and offer a robust 50 Gbps networking connection.
Google claims that N4A instances deliver "up to 2x better price-performance and 80% better performance-per-watt than comparable current-generation x86-based VMs." This effective cost reduction is partly due to Google’s in-house design of the Arm-based Axion processors, which allows them to source the components more affordably from manufacturers like Taiwan Semiconductor Manufacturing Company (TSMC). Research director at IDC, Brandon Hoff, emphasized that this allows Google to present a compelling alternative for enterprises.
Notably, Google is already leveraging Arm instances internally for services such as YouTube, Gmail, and BigQuery. This trend isn’t unique to Google; other cloud giants like AWS and Microsoft have also seen success with their Arm-based chips, Graviton and Cobalt, respectively, which have considerably lowered operational costs for internal workloads.
The Google N4A instances will be integrated across various services, including Compute Engine for virtual machines, Google Kubernetes Engine for containerized workloads, and Dataproc for big data analytics. Initially, users can access these instances in select regions, including parts of the US and Europe.
Furthermore, Google is set to release C4A instances, designed for heavier workloads like high-traffic application servers and AI-driven tasks. Additionally, a bare-metal instance known as C4A Metal will be available soon for specialized workloads in non-virtualized environments.
For more details, visit Google’s N4A Series.
