docs update

docs/benchmark/README.md

@@ -0,0 +1,43 @@
+# HTTP Hosts Benchmarking
+
+This repository contains tools and scripts for benchmarking HTTP hosts performance.
+
+## Prerequisites
+
+Before running the benchmarks, you need to create a path list (`path_list_500k.txt`). You have two options:
+
+1. Generate from real-world server logs using `nginx_to_path_list.py`
+2. Generate randomly (Note: real-world usage patterns are typically non-random, e.g., ocean tiles are rarely accessed)
+
+## Important Notes
+
+- Run the benchmarks on `localhost`, and not over the internet! Otherwise you'd be just testing your internet speed.
+- The benchmark uses [wrk](https://github.com/wg/wrk) HTTP benchmarking tool
+
+## Usage
+
+Basic command:
+```bash
+wrk -c10 -t4 -d10s -s /data/ofm/benchmark/wrk_custom_list.lua http://localhost
+```
+
+### Parameters Explained
+
+- `-c10`: Number of connections to keep open
+- `-t4`: Number of threads to use
+- `-d10s`: Duration of the test (10 seconds)
+- `-s`: Script file to use
+
+### Thread Count Considerations
+
+- `-t1`: More accurate results as the URL list is loaded exactly in sequence
+- `-t4`: Better reflects real-world usage patterns
+
+## Results
+
+Benchmark results can be found in [results.md](results.md)
+
+## Contributing
+
+Feel free to submit your results including which hosts were used.
+

docs/benchmark/nginx_to_path_list.py

@@ -0,0 +1,31 @@
+import json
+
+
+# This script parses a nginx server log and creates a text file
+# which can be used in the Lua script.
+# The path file is not suppied in this repo.
+
+with open('access.jsonl') as fp:
+    json_lines = fp.readlines()
+
+paths = []
+for i, line in enumerate(json_lines):
+    log_data = json.loads(line)
+    if log_data['status'] != 200:
+        continue
+
+    if log_data['request_method'] != 'GET':
+        continue
+
+    uri = log_data['uri']
+
+    if 'tiles/' not in uri or not uri.endswith('.pbf'):
+        continue
+
+    path = log_data['uri'].split('tiles/')[1]
+    paths.append(path + '\n')
+
+    print(f'{i / len(json_lines) * 100:.1f}%')
+
+with open('path_list.txt', 'w') as fp:
+    fp.writelines(paths)

docs/benchmark/wrk_custom_list.lua

@@ -0,0 +1,39 @@
+local counter = 1
+local lines = {}
+local url_base = "/planet/fake_version/"  -- trailing slash
+local path_list_txt = "/data/ofm/benchmark/path_list_500k.txt"
+
+for line in io.lines(path_list_txt) do
+    table.insert(lines, url_base .. line)
+end
+
+local function getNextUrl()
+    -- Get the next URL from the list
+    local url_path = lines[counter]
+    counter = counter + 1
+
+    -- If we've gone past the end of the list, wrap around to the start
+    if counter > #lines then
+        counter = 1
+    end
+
+    return url_path
+end
+
+request = function()
+    -- Return the request object with the current URL path
+    path = getNextUrl()
+    local headers = {}
+    headers["Host"] = "ofm"
+    return wrk.format('GET', path, headers, nil)
+end
+
+response = function(status)
+    if status ~= 200 then
+        print("Non-200 response")
+        print("Status: ", status)
+        -- this only works in single threaded mode (-t1)
+        print("Request path: ", path)
+    end
+end
+