Skip to content

Primatomic

High-performance distributed synchronization platform with distributed locks, key/value storage, and memory regions. Built on Raft consensus for enterprise reliability.

🔒 Distributed Locks

Create mutex or read/write locks with timeouts and automatic deadlock detection

🔑 Key/Value Store

Create namespaces, manage keys with TTLs, and perform atomic compare-and-swap operations

🚀 Memory Regions

Allocate large byte arrays up to 128GB with NVMe-oF backend for high throughput

⚡ High Performance

P99 latencies under 10ms for key/value ops, 20ms for locks, 5ms for memory I/O

All operations maintain linearizable consistency via Raft consensus.

  • Key/Value ops: P99 < 10ms
  • Lock operations: P99 < 20ms
  • Memory I/O: P99 < 5ms (NVMe-oF)
  • Linearizable operations via Raft consensus
  • Strong consistency across all operations
  • Fault tolerance with automatic leader election
  • Atomic compare-and-swap semantics
  • Multi-tenant namespace isolation

Start with our fully-managed cloud service:

Terminal window
# Sign up for free Developer tier (1 GiB RAM, 1M ops)
pip install primatomic

Note: Self-hosted deployment options will be available in future releases.

import primatomic
# Connect to Primatomic Cloud
client = primatomic.Client("your-workspace.primatomic.com:443",
jwt_token="your-api-token")
# Key-value operations with linearizable consistency
client.put_key("app", "user:123", b'{"name": "Alice"}')
user_data = client.get_key("app", "user:123")
# Distributed coordination
with client.lock("app", "payment-processing"):
# Critical section - only one service can execute this
process_payment(user_id)
# High-performance memory regions
client.allocate_memory("cache", "shared-buffer", 1024*1024)
client.write_memory("cache", "shared-buffer", 0, b"fast access data")
TierMonthly FeeRAM IncludedOps Included
DeveloperFree1 GiB1M
Starter$4910 GiB10M
Growth$19950 GiB50M
Scale$799256 GiB250M
EnterpriseCustom≥1 TiB≥1B

Memory Regions: Large byte arrays (up to 128GB) backed by NVMe-oF with direct I/O for high throughput

Operations: Billable API calls including key/value ops, lock acquire/release, and memory I/O

Namespace: Multi-tenant isolation boundary with separate data, locks, and billing

Raft Consensus: Distributed algorithm ensuring strong consistency and fault tolerance across cluster nodes

Ready to build? Start with our free tier →