This simple Worker was created to compare the response times between Cloudflare KV and D1. Since Cloudflare doesn't seem to offer any emperical analysis between the two, this repo at least gives a simple comparision between to two. Keep in mind, this is a very simple comparision, using only a single zone (assumedly).
The results seem to show that D1 is at least as fast as KV, and in some cases faster. This doesn't take into account cross-region propagation delays, since I don't know of a good way to test that easily.
I'd refrain from drawing any other big conclusions from this data, this is a very simple test.
insert_and_propagate tests will insert a value, then query within a loop until the newly inserted value is returned
| Metric | Description | Time |
|---|---|---|
| kv_insert | Insert value into KV | 138ms |
| kv_read | Read existing value from KV | 17ms |
| kv_insert_and_propagate | Insert and read value back from KV | 105ms |
| d1_insert | Insert value into D1 | 45ms |
| d1_read | Read existing value from D1 | 16ms |
| d1_insert_and_propagate | Insert and read value back from D1 | 66ms |
These times seem to vary quite a bit depending on the geo-location of the user hitting the end point, so I would recommend conducting your own tests before blinding accepting these results.
You'll need the following created in Cloudflare:
- D1 Database
- KV Namespace
Fork this repo, and update the wrangler.toml file with the configuration for your D1 and KV namespaces.
After that, simply running npm run deploy will deploy the worker to your Cloudflare account and setup the bindings. From there, simply invoke
the Worker to get the benchmarks.