Fast disks, attached to your VMs.
Resize live, snapshot continuously.
Persistent SSD-backed volumes for databases, OS disks, and any workload that needs sub-millisecond latency. Triple-replicated within the region. Resize without downtime.
What will it cost? Drag the sliders.
Block storage pricing is genuinely simple — capacity × tier rate, plus snapshots if you want them. No retrieval fees, no egress (volumes are private to your VM).
Min 10 GB, max 16 TB per volume. Attach up to 16 volumes per VM for more.
Snapshots are incremental — you only pay for the changed blocks. We estimate ~3 % of volume size per day for typical OLTP workloads.
$0.005/GB per copy. Most teams enable one for DR; multi-region copies are common in finance.
Three tiers, real numbers.
| Tier | Price | Baseline IOPS | Burst IOPS | Throughput | Latency p99 | Snapshot rate |
|---|---|---|---|---|---|---|
| Standard | $0.11/GB | 5,000 | 10,000 | 250 MB/s | < 5 ms | $0.04/GB |
| Performance | $0.18/GB | 10,000 | 30,000 | 750 MB/s | < 2 ms | $0.04/GB |
| Extreme | $0.30/GB | 40,000 | 80,000 | 2 GB/s | < 1 ms | $0.04/GB |
IOPS measured with 4 KB random reads, queue depth 32. Throughput at 1 MB sequential. Baseline is sustained; burst can be held for 30 minutes per hour.
When teams reach for block.
Random-read-heavy with WAL fsync on the critical path. Performance tier hits the latency floor without the cost of Extreme.
OS disks see almost no real I/O after boot. Standard tier is plenty — and the lowest $/GB on the menu.
Kafka brokers, Redis with persistence, write-heavy MySQL clusters. Sustained high IOPS without bursting.