ReleaseBytes
gcp Google Cloud Blog ·

Google Cloud C4N instances achieve GA for optimized network and storage I/O

bloggcpgapreviewengineermediagcp-gkegcp-cloud-storagegcp-dataflowgcp-compute-engine
feature patch

Google Cloud's C4N Compute Engine instances are now generally available, offering enhanced network and block storage performance for demanding enterprise workloads. These instances are purpose-built to eliminate I/O bottlenecks, providing up to 400 Gbps of network bandwidth and 25 GiB/s of storage throughput. This makes them ideal for network-intensive applications, large-scale databases, and AI/ML inference, delivering significant total cost of ownership (TCO) benefits.

  • →C4N instances offer high-performance network and block storage I/O
  • →Enhanced VM-to-VM and VM-to-internet network performance
  • →Optimized I/O for smaller VM shapes and improved Cloud Storage transfers
  • →Dynamic storage performance with Hyperdisk integration
  • →C4N instances provide significant performance gains for real-world applications
Features (1) ›
  • C4N instances offer high-performance network and block storage I/O

    C4N instances are now generally available, providing optimized network and block storage performance. They achieve up to 400 Gbps of network bandwidth and 25 GiB/s of storage throughput, leveraging Titanium offload architecture for dedicated hardware acceleration.

Enhancements (3) ›
  • Enhanced VM-to-VM and VM-to-internet network performance

    C4N instances deliver up to 400 Gbps of VM-to-VM network bandwidth and up to 200 Gbps of internet egress bandwidth. They also significantly increase packet processing performance, with up to 48 MPPS for internet egress, enabling non-blocking data delivery.

  • Optimized I/O for smaller VM shapes and improved Cloud Storage transfers

    Smaller C4N VM shapes (2-16 vCPUs) now offer predictable network bandwidth of 25–50 Gbps, reducing costs for I/O-bound tasks. Additionally, C4N VMs provide up to a 2x increase in bandwidth for Cloud Storage transfers, benefiting analytics and AI/ML workloads.

  • Dynamic storage performance with Hyperdisk integration

    When paired with Hyperdisk, C4N instances offer high block storage performance, with Hyperdisk Extreme providing up to 25 GiB/s throughput and 1M IOPS. This combination is now available across all C4N machine sizes, including smaller 2 vCPU configurations.

Notes (1) ›
  • C4N instances provide significant performance gains for real-world applications

    Real-world applications benefit from C4N's optimizations, with web serving seeing up to 1.5x more Nginx requests per second and databases like MySQL showing up to 45% better queries per second when data resides on disk.

Read the original announcement →

https://cloud.google.com/blog/products/compute/c4n-network-and-storage-optimized-vms/

Related releases