Linode Backup Service

What Linode Backup Service actually is

The Linode Backups service is a paid add-on for Akamai Cloud (Linode) instances, and it is deliberately simple: enable it with one click, and the platform starts taking automatic full-disk snapshots of your instance's disks on a schedule you choose. Up to four backups are kept — three generated automatically and one manual slot you control. The automatic slots are a daily backup (less than 24 hours old), a weekly backup (less than 7 days old), and a biweekly backup (between 8 and 14 days old), each taken inside a two-hour window you pick while the instance is still running. The fourth slot is a manual snapshot that persists until you overwrite it with another one.

The mechanics matter more than the marketing. Each backup is a full file-based snapshot of your disks, taken live — no downtime, no reboot, nothing to install. Backups are stored on separate dedicated hardware within the same data center as the instance, so a disk or host failure does not take your backups with it, but a data-center-level event plausibly could. Everything is manageable from Cloud Manager, the Linode CLI, or the REST API, which means you can script enabling backups across a fleet and query backup status from monitoring.

The restore workflow deserves a little detail, because it is where the paid add-on earns its keep over a hand-rolled cron job. Cloud Manager exposes the backup slots with their timestamps, and a restore can rebuild the instance's disks from a chosen backup while the instance configuration — IP addresses, plan, attached networking — stays intact. That means the common incident, a deployment that corrupted the working set, is a few clicks and a reboot rather than a rebuild-from-provisioning exercise. The manual snapshot slot adds the piece automation cannot: capture a known-good state immediately before a risky migration, keep it indefinitely, and roll back to it long after the automatic slots have rotated past.

Against a do-it-yourself backup to object storage, the trade is convenience versus control. Rolling your own gets you arbitrary retention, off-site copies, and per-file tooling like restic or borg; what it costs you is engineering time, monitoring, and the discipline to test restores. For one or two production instances, that engineering time dwarfs the price of this add-on, which is why we generally recommend paying for it even when a second layer exists.

There is one licensing detail worth knowing: if you buy Linode Managed — the operations service at $100 per instance per month — the Backups service is added to every Linode on the account at no extra charge. For Managed customers this is a bundled feature; for everyone else it is metered like everything else.

Pricing: what you really pay

The add-on starts at $2 per month for a 1 GB Shared CPU Linode (the Nanode class), and the price scales with instance size from there. Published North American rates run $2 for 1 GB shared, $2.50 for 2 GB, $5 for 4 GB, and $10 for 8 GB; as a rule of thumb across the catalog, expect roughly 20% of the instance's monthly price — the same ballpark as Hetzner's and DigitalOcean's backup add-ons. Billing is hourly with a monthly cap, consistent with how the rest of the platform charges.

Read against the competition, the pricing is fair but the value depends entirely on how much retention you need. Hetzner also charges about 20% and keeps seven daily restore points. DigitalOcean's basic weekly option keeps four weekly points at +20%, with a +30% daily variant. Vultr's automatic backups keep only the latest two scheduled points. Linode gives you four slots total — daily, weekly, biweekly, and a manual snapshot — for roughly the same money. It is a defensible middle: more frequent than weekly-only plans, far shallower than seven-daily retention.

The hidden cost is coverage. Backups are billed per Linode, and only for instances in core compute regions — the service is not available in distributed compute regions at all. If you run twenty instances, that is twenty add-on line items, and the number grows linearly with your fleet. At $2 for the smallest instance the entry cost is trivial; at scale, compare it against rolling your own to object storage before assuming the add-on is cheaper.

Performance & infrastructure

The service is file-based, not block-based, and it operates on unencrypted ext3 or ext4 file systems only. Those two constraints decide whether it works for you. Encrypted disks are out. Attached Volumes are not backed up as part of the service — only the instance's own disks. And backups are deleted along with the Linode: if you delete the instance, its backups go with it, which makes this a restore mechanism for live infrastructure, not an archive.

Retention is the other hard limit. Automated backups are never kept longer than 14 days — the oldest automated slot is the biweekly one, between 8 and 14 days old. Only the manual snapshot slot has unlimited age, and it holds exactly one image. In practice that means your longest automated recovery window is about two weeks, and long-term recovery depends on you deliberately saving manual snapshots, one per instance, or exporting data elsewhere.

Within those limits, the operation is well built. Backups are taken while the instance runs, so there is no maintenance window to schedule; restores can bring back a full system to the instance or a replacement, which is exactly what you want after a botched upgrade or a bad deploy. The service is covered by Akamai's non-compute SLA at 99.9% monthly uptime — lower than the 99.99% compute guarantee, but with the same pro-rata credit mechanism, claimable through support within 30 days. Documentation is explicit about every limitation above, which we count as a positive: the product page lists its own exclusions instead of hiding them.

For anything beyond single-instance protection, the intended shape is a layered strategy — and the vendor's own documentation says as much, positioning the service as one tier of a multi-tier backup approach rather than a complete one.

Where it falls short

Retention is shallow. Four slots with a hard 14-day ceiling on automated history is thin for compliance, audit, or ransomware scenarios where the compromise is discovered weeks after the fact. Providers in the same price band keep seven daily restore points; Linode keeps one daily, one weekly, one biweekly.

The storage location is a structural weakness: same data center as the instance. The backups are off-host but not off-site, so a facility-level incident can take your production and your recovery copies together. For a genuine 3-2-1 strategy you need a copy somewhere else — object storage in another region, another provider, or an external system — and that copy is entirely your problem to build and test.

The coverage gaps add up: attached Volumes are excluded, unencrypted ext3/ext4 only, backups die with the instance, and distributed compute regions are unsupported. Each one is documented, and each one has quietly burned someone. If your disk layout includes encrypted volumes or you rely on Volumes for data, this service protects less of your stack than its name suggests.

Finally, there is no cross-instance management story: pricing is per instance, retention policy is per instance, and there is no portfolio-level reporting. Fleet operators will be scripting around the API sooner or later.

Who should pick this

We would pick Linode Backups as the low-effort safety net for individual Akamai Cloud instances — small production servers, staging boxes, and any workload where "restore the last daily backup after a bad deploy" covers 95% of realistic incidents. At $2 per month on an entry instance it is cheap insurance, the live-snapshot mechanics are convenient, and the API makes it scriptable. If you are already paying for Linode Managed, take it — it is included.

We would not pick it as the only layer of a backup strategy, for anything with compliance retention requirements, or for fleets where per-instance pricing and per-instance policy will not scale. In those cases, pair it — or replace it — with off-site object storage and your own retention policy; our backup-solutions category lists options that keep history for months rather than days. Used within its limits, the Backups service does exactly what it promises; the failure mode is expecting it to do more.

Key takeaways

• Paid add-on from $2/month (1 GB shared instance); roughly 20% of instance price at larger sizes; billed hourly with a monthly cap
• Four backup slots: daily (<24h), weekly (<7d), biweekly (8-14d), plus one manual snapshot that persists until overwritten
• Automated history is capped at 14 days — shallow retention is the main strategic limitation
• Backups are live file-based snapshots of unencrypted ext3/ext4 disks, stored off-host but in the same data center
• Excludes attached Volumes, encrypted disks, and distributed compute regions; backups are deleted with the instance
• Good convenience layer for single instances; pair with off-site storage for a real 3-2-1 strategy

Sources

• https://techdocs.akamai.com/cloud-computing/docs/backup-service
• https://www.akamai.com/legal/sla-non-compute
• https://linode.com/docs/products/services/managed/guides
• https://linode.com/community/questions/20257/linode-backup-pricing-questions
• https://stackvaluelab.com/vps-backup-pricing
• https://vpsbenchmarks.com/hosters/linode
• http://en.wikipedia.org/wiki/Linode