Deno KV — Live Read/Write Latency Benchmarks

This application compares read and write latencies across five serverless databases: Deno KV, Upstash Redis, AWS DynamoDB Global Tables, Firestore, and Cloudflare Workers KV.

We're able to minimize network latencies by using serverless functions hosted in the same region as the database.

We've inserted 14,275 records into each database. Each record is GitHub repo metadata obtained from the GitHub API. Each database has an index on the forks_count field.

Check out the source or read the blog post for more details.

Current Request

Loading...
!--frsh-currentrequest:0-->

Read Latency

— 50th percentile

Loading...!--frsh-readwritelatencies:1-->

— 90th percentile

Loading...!--frsh-readwritelatencies:2-->

— 99th percentile

Loading...!--frsh-readwritelatencies:3-->

Write Latency

— 50th percentile

Loading...!--frsh-readwritelatencies:4-->

— 90th percentile

Loading...!--frsh-readwritelatencies:5-->

— 99th percentile

Loading...!--frsh-readwritelatencies:6-->

* Due to Cloudflare KV's heavy caching and eventual consistency, it isn't well suited for this test at lower intervals. In this test Cloudflare KV is cached with a longer expiration time than the other services, which causes the smaller sample sizes.