One Year of Warm Migrations

Abe Massry

Jul 16, 2025

Abe Massry

Abe Massry

Written by

Abe Massry

Abe Massry is a Senior Software Engineer at Akamai, where he specializes in virtual machine lifecycle management within the Core Compute team. With deep expertise in cloud infrastructure, Abe contributes to architectural and engineering efforts across Linode and Akamai’s compute platforms.

Share

Why Observability Tools Tend to Fail at Scale

One year ago Akamai rolled out warm migrations to the Linode platform for both data center and host migrations, and resizing operations. During a traditional cold migration, the target Linode is shut down, its data copied to a new host, and the new Linode booted. A warm migration differs from a traditional cold migration in that the target Linode continues running until the Linode is synced to a new host. At that point, the Linode powers down, and because the local NVMe data is mirrored one-to-one on the new host, the new Linode boots exactly the same as it would a reboot on the previous host. If a resize is necessary, the resize can happen on the new host while the Linode is powered down and immediately booted back up.

We pioneered this new technology for two reasons, both crucial to combatting customer downtime, or the time a Linode is occupying space but is powered down. The first reason is that cold resizes (previously known as resizes) required the Linode to stay powered down for the duration of the migration; the resize would take place on the destination host, and then the new Linode would boot on the new host. A migration is needed for resizes because the current Linode’s host is not guaranteed to have space if the Linode is sized up, and if the Linode is sized down, better hardware on a new host might be available. For larger Linodes, this required anywhere between a few minutes to hours of downtime. The goal was to drastically shorten this timeframe. The second reason we developed warm migrations is that Live Migrations can only operate between compatible CPU types. A warm migration doesn’t have this restriction.

Customers started warm-migrating last year, and we’re proud to be able to share just how much downtime we actually saved. In a year, customers warm-migrated 11.6 Petabytes of data across over 58,000 Linodes, saving a total of more than 6,100 hours of downtime. Again, this downtime is the time that the Linodes were processing requests, serving customers, supporting business, and doing everything else an Akamai user needs, instead of being powered down, while still able to resize and migrate to new destinations.

Data Migrated 11.6 Petabytes
Migrations 58,000 Linodes
Downtime saved 6,100 Hours

It took an outsized effort to bring Warm Migrations to life, but we constantly strive to meaningfully improve the experience customers have on our platform. Multiple teams across the company worked together with a unified vision to see this project through to the end, and from an engineer’s standpoint it’s very rewarding to see this large impact. If you’d like to try warm migrations out, the feature is available as a resize option on the Cloud Manager. And, if working on large, impactful projects like this interests you, we’re hiring

Abe Massry

Jul 16, 2025

Abe Massry

Abe Massry

Written by

Abe Massry

Abe Massry is a Senior Software Engineer at Akamai, where he specializes in virtual machine lifecycle management within the Core Compute team. With deep expertise in cloud infrastructure, Abe contributes to architectural and engineering efforts across Linode and Akamai’s compute platforms.

Tags

Share

Related Blog Posts

Cloud
Akamai Cloud Is Built for What Cloud Has Become (Updated May 2026)
May 06, 2026
Explore Akamai Cloud’s Q1 2026 updates: New NVIDIA Blackwell GPUs, edge native Akamai Functions, and infrastructure built for real-time, distributed AI inference.
Cloud
AI Survey: 50% of Organizations Struggle to Maintain Latency at Scale
May 06, 2026
The Akamai State of AI Inference report captures real data from the field that describes how AI inference is being built and scaled in production today.
Cloud
AI Inference at the Edge: How Akamai’s Evolution Made It Possible
November 26, 2025
Read how Akamai, a pioneer in CDNs, is undergoing a profound strategic transformation into a diversified cloud and cybersecurity provider in the AI era.