Google Cloud announced general availability of the M4N machine series on Compute Engine, purpose-built for I/O-intensive, high-memory workloads. It is the second offering in the network- and block-storage optimized VM family, following the M4 series. M4N combines 5th Gen Intel Xeon Scalable processors with Google’s custom Titanium offload architecture, and the company claims it provides the highest per-core IOPS and throughput for high-memory instances among leading hyperscalers, along with over 20% TCO reduction for Oracle databases compared to similar offerings. It is also described as the industry’s first network- and block-storage-optimized instance with higher memory ratios (up to 26:1) and sizes (6TB).

M4N delivers up to 25,000 MiB/s (25 GiB/s) of aggregate host storage performance and up to 1 million IOPS when paired with Hyperdisk Extreme, doubling the block storage performance of current M4 instances. It offers up to 400 Gbps aggregate VM-to-VM network bandwidth and up to 50 Gbps single-flow bandwidth within the same VPC, plus up to 200 Gbps internet egress and 48 MPPS packet processing, all without requiring premium Tier-1 networking add-ons. Instances scale from 16 to 224 vCPUs with up to 5,952 GB of DDR5 RAM across three memory-to-vCPU ratio tiers, and support Resource-based Committed Use Discounts.

The family targets workloads that need both dense RAM and sustained I/O, such as mission-critical databases (Oracle, SAP HANA, SQL Server, IBM DB2, MySQL, PostgreSQL), generative AI and RAG data layers (Milvus, Pinecone, Qdrant, Vespa, Redis), enterprise healthcare and ERP systems (Epic Systems, SAP ECC, SAP S/4HANA), and real-time analytics/EDA workloads. The high memory-to-core ratio (up to 26.57 GB/vCPU) is intended to reduce Oracle licensing costs by avoiding over-provisioning of vCPUs, improving cache-hit ratios by keeping larger SGA regions in memory, and enabling right-sizing of infrastructure.

Early customer feedback highlights the elimination of over-provisioning. Sherri Trojan of Sabre noted that M4N delivers high throughput across smaller to larger shapes. Tessell’s CEO Bala Kuchibhotla praised the platform for enterprise Oracle deployments requiring scalability, resiliency, and performance. Intel emphasized the combination of 5th Gen Xeon and Titanium architecture for extreme memory capacity and I/O throughput.

With Hyperdisk storage, users can independently tune IOPS, throughput, and capacity. Hyperdisk Extreme provides up to 25 GiB/s and 1,000,000 IOPS; Hyperdisk Balanced scales to 20 GiB/s and 640,000 IOPS for cost-effective enterprise storage. M4N bridges the previously separate paradigms of large memory footprints and extreme I/O performance, minimizing bottlenecks via Titanium offloads without infrastructure add-ons. The instances are available in select regions, with full configuration details in the documentation.