Google Cloud unveiled the high performance C4N instance type at Google Next 2026. C4N is a network-optimized instance purpose-built for high-volume data traffic and I/O-sensitive workloads — agentic AI and real-time analytics  —  exactly the kind of infrastructure that could move the needle for Sycomp Intelligent Data Storage customers.

As a Google Cloud partner in the C4N preview program we ran the tests. Here's what we found.

Why the C4N Matters for HPC Storage

Most Compute Engine instances ship with a single high-speed network interface. The C4N doubles the bandwidth potential with two 200 GbE interfaces.

That architectural shift is significant. It means we can dedicate both networks entirely to storage traffic, doubling the available bandwidth for data-intensive workloads, and achieving 2x storage performance over the previous generation.  On behalf of our storage and HPC customers, we verified those numbers against a real-world IO500 benchmark.

At Sycomp, we are inspired by our customers’ real-world applications and challenges, and at the same time move at the pace of innovation. The C4N was announced just weeks ago and is already active in our test environment, ensuring our customers can evaluate the latest GCP capabilities without delay.

Test Configuration

To ensure conservative, reproducible results, testing was conducted using the smallest production configuration Sycomp offers with three storage nodes. This represents the baseline of the platform’s capabilities, not its upper limit. The architecture scales to 128 nodes — approximately 42x the capacity of the environment tested here — and the results should be viewed within that broader scalability context.

• 3-Storage nodes (minimum available configuration)
• Benchmark: IO500 version io500-isc26 (standard)
• No custom tuning applied (out-of-the-box results only)
• Google Cloud C4N instance series, dual high-speed network interfaces

Results

58.5 GB/s Peak sequential read	58.6 GB/s Peak sequential write	1.2M Files/sec (find)

Google Cloud’s published maximum for the C4N is 60 GB/s. In our tests, Sycomp achieved 58.5 GB/s on read and 58.6 GB/s on write — 97% of the theoretical ceiling with zero platform-specific tuning.

That's a strong starting point, and there's measurable room to close the remaining gap through configuration work we haven't yet done.

"97% of the theoretical 60 GB/s ceiling, out of the box, on our minimum node count. The C4N isn't just faster — it changes the price-performance equation for storage workloads on Google Cloud."

The metadata story is equally compelling. The find benchmark returned 1,217,553 kIOPS — over 1.2 million file-per-second operations. For workloads that are heavy on small file access, directory traversal, or metadata operations (think AI training pipelines, large-scale genomics, or financial tick data), this is a meaningful result.

Full IO500 Benchmark Breakdown

Test	Result	Time (sec)
IOR easy write	58.55 GiB/s	308.4
IOR easy read	58.51 GiB/s	308.6
IOR hard write	21.84 GiB/s	300.5
IOR hard read	48.76 GiB/s	134.6
IOR random 4K easy read	0.70 GiB/s	304.1
MDtest easy write	34.03 kIOPS	932.7
MDtest easy stat	165.86 kIOPS	192.0
MDtest easy delete	57.50 kIOPS	568.3
MDtest hard write	8.58 kIOPS	423.9
MDtest hard stat	165.80 kIOPS	22.9
MDtest hard read	62.73 kIOPS	58.8
MDtest hard delete	2.08 kIOPS	1746.6
Find	1,217.55 kIOPS	29.0

What This Means for You

The C4N delivers two things that matter to storage customers: more throughput and a better price-to-performance ratio.

The dual-network architecture means we're not fighting over bandwidth between compute and storage traffic, a constraint that quietly taxes performance on conventional single-NIC instances. And because this is our minimum configuration, any customer who needs to scale can do so linearly, up to 128 nodes.

We'll be fine-tuning as we continue testing, but even in this baseline state, the C4N sets a new bar for what Google Cloud-based HPC storage can deliver. The best news is, Sycomp is ready to provision it for customers today.