LavikFrom performance to economics
Reduce the cost of growing Redis datasets.
Put growing value payloads on NVMe SSD and reduce the need for expensive DRAM. Enter your prices to calculate what that means for your dataset.
What does a 20:1 capacity-price ratio mean?
If DRAM costs 20 times as much per GiB as SSD, placing the same value payload on SSD costs one twentieth as much for that storage component: 95% less. This is arithmetic under a stated price assumption, not a hardware quotation or a measured whole-deployment saving.
Lavik still needs memory for indexes and runtime state. CPU, replicas, usable capacity, and operations also enter a complete comparison. The model below shows how those factors affect the overall ratio.
A worked example: 1 TiB of value payload
Assume DRAM and NVMe SSD capacity prices of 20 and 1 cost units per GiB. One TiB is 1,024 GiB: value capacity alone costs 20,480 units in DRAM or 1,024 units on NVMe SSD, saving 19,456 units (95%). These are illustrative units, not hardware quotes; add index memory, replicas, and shared server costs separately.
Use the model below to see how index memory and shared costs change the deployment cost ratio.
Adjustable assumptions
Defaults illustrate the formula; they are not hardware quotes or measured memory usage. Assumes equal replica counts, no compression, and storage amplification of 1.
84.5% lower modeled Lavik cost
Lavik = shared + index × DRAM + payload × SSD
This model does not predict latency, durability, or availability. Performance requirements may change the required server count and configuration.
Throughput also determines node count
Capacity is one part of the budget. For a fixed dataset and QPS target, higher per-node throughput can reduce the nodes added to serve traffic. The comparison calculators combine measurements at matched concurrency with capacity, throughput headroom, replicas and node prices.
Compare costs by product →