Linode is rolling out faster, better performing, and more reliable NVMe (non-volatile memory express) storage across its entire block storage fleet. The rollout kicks off in the company’s Atlanta data center, with the remainder of its global network upgrading over the next quarter.
Linode’s NVMe rollout includes the company’s first erasure-coded block storage cluster, a method for storing data in Ceph that uses an algorithm to slice data into chunks, resulting in significant improvement in storage efficiency over replicated clusters.
“We put a lot of effort into making our block storage platform more performant and efficient,” said Chris Aker, founder and CEO at Linode. “Using erasure-coded clusters, we’re able to extract 160% more usable storage out of the same raw disk capacity. It’s an incredible increase in storage efficiency done in a way that allows us to deliver high performance without creating a premium tier like most other providers would.”
NVMe is the hardware interface for next-generation SSDs (solid-state devices). The technology represents a significant improvement over traditional spinning hard disk drives, with a 10-20x increase in throughput and up to 2000x improvement in IOPS. Linode is one of the first hyperscale cloud providers to deploy NVMe at scale.
NVMe Block Storage makes attaching highly available and scalable storage volumes to cloud instances easy and ultra-fast. With NVMe-backed block storage, customers can experience a significant performance upgrade to power use-cases such as database storage and other tasks where high storage speeds and consistency are critical.
“Making things faster is good. Making them faster, more resilient, and more reliable is even better. And doing it in a way that doesn’t cost customers more is what developers have come to expect from us,” said Aker.