Self-managed cloud compute decision guide
Akamai Cloud Computing is the current portfolio name; official documentation and Cloud Manager still call a virtual machine a Linode. The useful buying question is whether the compute family, deployment surface, transfer model and recovery boundary fit your application and operating team.
Evidence checked 20 August 2026. Prices, regions and service availability can change. No lab benchmark, ticket test, uptime measurement, user survey or paid ranking is claimed.
Decision first
Akamai documents Shared CPU for workloads that do not need sustained compute availability. Short bursts can reach 100%, but average sustained use should remain below 80%, including virtualization overhead. That is operating guidance—not a benchmark or promise of application speed.
As of 20 August, the documented range is 1–192 GB memory, 1–32 shared virtual CPUs and 25–3,840 GB storage, starting at US$5/month or US$0.0075/hour. Use for development, staging and lower-traffic workloads that tolerate shared-core contention.
The documented range is 4–512 GB memory, 2–256 dedicated virtual CPUs and 40–7,200 GB storage, starting at US$36/month or US$0.05/hour, with regional variation. Prefer it when contention or sustained CPU affects the service.
The documented range is 24–300 GB memory and 2–16 dedicated virtual CPUs, starting at US$60/month or US$0.09/hour. It fits in-memory data or cache workloads; it is not a general-purpose upgrade recommendation.
Progressive decision aid
The result is guidance, not a quote. The same fallback rules remain readable below when JavaScript is unavailable.
Deployment surface
Distributed regions are limited-availability environments for edge-native workloads. Access requires an existing account with a core-region deployment and contact with Akamai. Applications should tolerate the loss of one region or server.
| Capability | Core regions | Distributed regions | Decision |
|---|---|---|---|
| Compute family | Shared, Dedicated, High Memory and specialized families | Dedicated CPU only | Prove workload and price fit again |
| Block/Object Storage and Backups | Available subject to region/product checks | Unavailable directly | Stateful designs need a core dependency or another recovery design |
| VPC | Available subject to current region support | Unavailable since 22 April 2026 | Do not carry a VPC-dependent design unchanged |
| Kubernetes, NodeBalancers and managed databases | Available subject to current region support | Unavailable directly | Verify every dependency, not just the city |
| Migration | Core-to-core workflows available | Cold/warm between distributed regions; no live or cross-surface migration | Keep an application-level migration path |
| Access | Normal account availability subject to capacity | Limited; existing core use plus contact required | Do not promise instant deployment |
Current distributed locations listed by Akamai: Auckland, Berlin, Bogotá, Denver, Hamburg, Houston, Johannesburg, Kuala Lumpur, Marseille, Querétaro and Santiago. Treat the list as dated inventory, not permanent availability.
Network cost
The current pricing page lists core-region egress overage at US$0.005/GB and distributed-region egress at US$0.01/GB. The transfer documentation separately lists regional exceptions, including Jakarta and São Paulo. Recheck the exact region and plan before purchase.
Never assume: one global pool, universal included bandwidth or a transfer-free specialized plan.
Complete cost
Representative amounts above were verified on 20 August 2026 and are not a guaranteed quote. Regional pricing and plan generations can differ.
Recovery
| Question | Documented answer | Required evidence |
|---|---|---|
| How many copies? | Daily, weekly, biweekly and one manual snapshot; automated copies are no older than 14 days | Record successful job times and retention |
| Where? | Separate hardware in the same data centre | Maintain an encrypted offsite copy |
| What is excluded? | Attached Volumes, configuration profiles, access-control lists and extended attributes | Inventory and back up every excluded surface |
| Database consistency? | A running transaction can leave database files unclean | Schedule application-aware database dumps |
| Portability? | Restore to a Linode in the same data centre, then transfer normally | Test an isolated restore and export |
| After deletion? | Deleting the Linode deletes its backups | Prove independent copies before deletion |
Responsibility
| Layer | Primary owner | Proof before production |
|---|---|---|
| Cloud platform and documented service incidents | Akamai | Status path, ticket route and service-specific documentation |
| Linux, SSH, firewall and packages | Customer | Hardening, patch cadence, least privilege and logs |
| Web stack, database and application | Customer | Monitoring, security updates, data integrity and incident runbook |
| Backups and recovery design | Customer | Coverage map, offsite copy, database dump and restore test |
| Third-party software troubleshooting | Customer/vendor | Escalation ownership and paid expertise if required |
No SLA percentage is published here: the current legal scope, exclusions and claim mechanism were not independently verified to the standard required for a consequential uptime claim. Evaluate any service commitment directly before purchase.
Safe deletion
Use the migration acceptance and rollback process · Follow the evidence-led cancellation sequence · Design independent recovery
Alternatives
Compare Droplet responsibility, regional service coverage, current backup model and complete cost.
Compare exact plan/region fit, stopped-instance billing, backup retention and exit controls.
Compare network-zone cost, processor architecture, IP resources and Volume recovery boundaries.
If nobody owns Linux and application operations, compare a managed application platform instead. Use the true-cost framework, support-scope guide, location method, payment checks, comparator and migration path for the wider decision.
Evidence trail
Provider facts are drawn from current Akamai pricing and technical documentation. External links are direct and untracked.
Questions
Yes. Akamai completed its acquisition of Linode in March 2022. Current documentation uses Akamai Cloud Computing for the portfolio and Linode for a compute instance.
No. A Linode is self-managed compute infrastructure. The customer owns operating-system, application, monitoring, security and recovery work unless a separate managed service is purchased.
No. Powered-off Linodes remain billable because their data and reserved resources remain allocated. Deletion is required to stop the instance charge.
The paid Backups service stores up to four backups: daily, weekly, biweekly and one manual snapshot. Automated backups are no more than 14 days old.
No. Attached Block Storage Volumes and configuration profiles are excluded, and database files can be captured in an unclean state without application-aware dumps.
No verified Akamai Cloud or Linode affiliate relationship was identified for this page. Provider links are direct and untracked.
