LavikCapacity on compact instances
From one million to 100 million keys.
Read the report →LAVIK 0.2.0 / USE CASES
Keep more useful data online.Serve it at the speed your application needs.
Keep a larger catalog, longer-lived sessions and more features per entity. Lavik stores values on NVMe SSD and serves them through a Redis-compatible interface, so retaining more data no longer means buying DRAM for every byte.
1Bkeys × 1 KiB values
652,881GET QPS at 320 connections
0.943 msGET p99.9 at that same point
Lavik 0.2.0 · October 21, 2026 · 16 vCPUs · six NVMe SSDs. Full benchmark and reproduction →
BY WORKLOAD
Room for growing caches and online state.
Large application caches
Keep the long tail without a DRAM-sized bill.
Retained object payload and expensive origin missesOnline feature serving
Serve more entities and feature versions.
Entity cardinality × feature width × model versionsSessions & user profiles
Retain more user context across devices.
Retained context per user, device, and tenantReal-time leaderboards
Scale seasons, cohorts, and ranked state.
Members × boards × retained seasonsBY INDUSTRY
Put more data behind every interaction.
E-commerce & marketplaces
Catalog breadth without a memory-capacity tax.
Catalog breadth, seller offers, locales, and snapshot versionsAdtech & personalization
Broader decision context within a latency budget.
Reachable identities × context width × retentionGaming & live services
Keep player state and past seasons accessible.
Registered players, game modes, regions, and retained seasons20×
lower value-capacity cost
At a 20:1 DRAM/NVMe SSD price per GiB, the same value capacity costs 95% less in media. Plan the complete budget with your actual prices, index memory and replicas.
Calculate your capacity cost →From a compact instance to a billion keys.
On a 2-vCPU, 16 GiB instance, Lavik holds 100 million 1 KiB values while retaining nearly all of its smaller-dataset peak GET throughput. In the 16-vCPU test, its peak GET and SET throughput exceeds the tested Redis and Valkey configurations in DRAM. Choose the report closest to your application’s scale.
Compare Redis and Valkey
NVMe SSD throughput against DRAM.
Read the report →Compare SSD-backed databases
Matched concurrency at one billion keys.
Read the report →Put your first dataset on Lavik.
Install 0.2.0, connect your existing client, then rehearse migration with RedisShake.