High-performance NVMe block volumes up to 16 TB that stay independent of compute. Attach, detach, and hot‑swap volumes across virtual machines without rebooting — ideal for databases, media, backups, and even bootable OS volumes.
Why choose Akamai Block Storage
NVMe speed for demanding workloads
10x–20x more throughput and up to 2,000x the IOPS of spinning disks.
Consistent low‑latency performance for databases, logs, and transactional apps.
Persistent, independent volumes
Volumes are provisioned separately from compute and persist even if VMs are deleted. See the Block Storage TechDocs for details and limitations.
Hot‑swappable portability
Move volumes between instances in the same data center without downtime or rebooting.
Built‑in resiliency
Data durability with fault‑tolerant replication and erasure coding within the storage cluster.
Scale on demand
Add or resize volumes up to 16 TB per volume as your needs change — without disrupting running workloads.
Open tooling and automation
Full API, CLI, and Terraform support for lifecycle management.
Works with your storage mix
Pair Block Storage with S3‑compatible Object Storage for data lakes, archives, and large unstructured datasets.
Pricing
Simple, transparent pricing starting at $0.10 per GB per month ($0.00015/GB per hour), from 10 GB to 16 TB.
How it works: durability, fault tolerance, and persistence
Independent of compute: Volumes are created and managed separately from VMs and remain available until you explicitly delete them.
Fault tolerance: Data is stored with built‑in replication/erasure coding inside the storage cluster, helping protect against hardware failures.
Hot‑swap operations: Detach from one VM and attach to another VM in the same data center live, enabling migrations, blue/green swaps, and maintenance without downtime.
NVMe‑backed media delivers high IOPS and throughput suitable for OLTP databases, time‑series systems, analytics scratch space, and CI/CD artifact stores.
Use multiple volumes to stripe for higher aggregate throughput, or separate logs and data files for predictable latency.
Complement hot data on Block Storage with cold/warm data in S3‑compatible Object Storage.
Akamai Block Storage vs. AWS EBS (IOPS, price/GB, portability)
IOPS performance
Akamai: NVMe‑backed volumes designed for high IOPS and low latency; performance is included with capacity.
AWS EBS: Performance varies by type (e.g., gp3, io2). You can provision IOPS and throughput (often at additional cost).
Pricing per GB
Akamai: Starts at $0.10/GB‑month with straightforward, per‑GB pricing.
AWS EBS: Price depends on volume type and region (for example, gp3 is commonly priced lower per GB in some regions, with separate IOPS/throughput charges). Always check current regional pricing.
Volume portability
Akamai: Hot‑swap volumes between VMs in the same data center without rebooting; volumes persist independently of VM lifecycle.
AWS EBS: Detach/attach between instances within the same Availability Zone; portability across AZs/regions typically uses snapshots.
Bottom line: If you want simple per‑GB pricing with NVMe performance and hot‑swappable persistence, Akamai Block Storage is a strong fit. If you need granular, provisioned IOPS with advanced cross‑AZ options, EBS offers many tiers but with more pricing complexity.
How Akamai NetStorage supports large media and static assets
Akamai NetStorage is origin storage integrated with Akamai’s global delivery platform, built for large files, VOD libraries, software downloads, and static site assets.
Distributed origin: Content is stored across multiple storage locations for durability and availability.
Tight CDN integration: Akamai edge servers fetch from NetStorage, minimizing origin load and improving first‑byte times at global scale.
Ingest and access: Supports high‑throughput ingest and origin fetch via standard protocols/APIs; ideal for workflows that need rapid publication and global fan‑out.
Use case fit: Pair NetStorage with Akamai media delivery or download delivery to serve large libraries with predictable performance and cost.
Use Block Storage for hot, VM‑attached workloads (e.g., databases, transcoders’ working sets) and NetStorage or Object Storage for origin/static distribution and long‑lived assets.
Reference architectures on Akamai Cloud
Transactional and operational workloads (databases, line‑of‑business apps)
Compute: CPU instances or Managed Databases.
Storage: Akamai Block Storage for database files, logs, and temp; snapshots/backups to Object Storage.
Networking: Private Networking for low‑latency east‑west traffic; Cloud Firewall for segmentation; NodeBalancers for HA.
Operations: Hot‑swap volumes to refresh or replace VMs without downtime.
Analytics and machine learning
Data lake: S3‑compatible Object Storage for raw/curated data.
Compute: GPU and CPU nodes pull “hot” working sets to Block Storage NVMe volumes for fast ETL/feature prep/training scratch.
Pipelines: Write checkpoints and artifacts back to Object Storage for durability and sharing.
Media and static delivery at scale
Origin: Akamai NetStorage or Object Storage for catalogs.
Processing: Transcoding/packaging jobs use Block Storage for fast, local I/O.
Delivery: Akamai CDN services front the origin for global, cache‑efficient delivery.
How to choose block storage for hot‑swapping and zero downtime
Prioritize:
- Live attach/detach without rebooting and consistent device naming.
- Independent lifecycle from compute; volumes must persist if VMs are deleted.
- Durability via replication/erasure coding and fast recovery from host failures.
- Clear per‑GB pricing and no surprise IOPS/throughput charges.
- Automation support (API, CLI, Terraform) for safe, repeatable swaps.
Akamai Block Storage checks these boxes with hot‑swappable, persistent NVMe volumes managed independently of compute.
What to evaluate for NVMe storage when migrating databases
IOPS and latency: Low single‑digit millisecond latency and sustained IOPS for your workload profile (random vs. sequential).
Throughput ceilings: Ensure per‑volume and per‑instance throughput meets peak demand; consider striping if needed.