Add the rest of the datasets

README.md

@@ -0,0 +1,3 @@
+# Read-only benchmark
+
+This tests the latency, sequential read, and random read speeds of a variety of data.

src/main.rs

@@ -1,8 +1,9 @@
-use curl::easy::{self, Easy as easy_curl};
+use curl::easy::Easy as easy_curl;
 use rand::{self, RngCore, SeedableRng};
 use rand_xorshift::XorShiftRng;
 use std::{
-    fs::{exists, File},
+    env::current_dir,
+    fs::{create_dir_all, exists, File},
     io::{Error, Write},
     os::unix::fs::FileExt,
     process::Command,
@@ -12,7 +13,7 @@ use std::{
 
 fn large_random_file_generation(path: String) {
     // https://stackoverflow.com/a/65235966
-    let mut out = Arc::new(Mutex::new(File::create(path)));
+    let out = Arc::new(Mutex::new(File::create(path)));
     // NOTE: update this both here and in the helper (_large_random_file_generation_helper())
     let num_threads: u64 = 12;
     let mut threads: Vec<JoinHandle<()>> = Vec::new();
@@ -49,61 +50,124 @@ fn _large_random_file_generation_helper(i: &u64, out: Arc<Mutex<Result<File, Err
         //let offset: u64 = (i * blocks_per_thread * 1310720) + (1310720 * u);
         let offset: u64 = u * block_size;
         let mut out = out.lock().unwrap();
-        out.as_mut().unwrap().write_all_at(&data, offset);
+        out.as_mut().unwrap().write_all_at(&data, offset).unwrap();
     }
 }
 
+/*
 fn single_threaded_large_random_file_generation(path: String) {
     let mut out = File::create(path).unwrap();
     let mut rng = XorShiftRng::seed_from_u64(2484345508);
     let mut data = [0u8; 65536];
-    for i in 0..409600 {
+    for _ in 0..409600 {
         rng.fill_bytes(&mut data);
-        out.write_all(&data);
+        out.write_all(&data).unwrap();
     }
 }
+*/
 
 fn small_random_files_generation(folder: String) {
     let mut rng = XorShiftRng::seed_from_u64(2484345508);
-    let mut data = [0u8; 1024];
+    let mut data: [u8; 1024] = [0u8; 1024];
     for i in 1..1001 {
         let mut out = File::create(format!("{folder}/{i}")).unwrap();
         rng.fill_bytes(&mut data);
-        out.write(&data);
+        out.write_all(&data).unwrap();
     }
 }
 
-fn grab_kernel(folder: String) {
+fn create_empty_file(path: String, size: u64) {
+    let out = File::create(path).unwrap();
+    out.write_all_at(&[0], size - 1).unwrap();
+}
+
+fn small_empty_files_generation(folder: String) {
+    for i in 1..1001 {
+        let out = File::create(format!("{folder}/{i}")).unwrap();
+        out.write_all_at(&[0], 1023).unwrap();
+    }
+}
+
+fn grab_kernel(folder: String, kernel_version: String) {
     // https://cdn.kernel.org/pub/linux/kernel/v6.x/linux-6.6.58.tar.xz
-    let mut curl = easy_curl::new();
-    curl.url("https://cdn.kernel.org/pub/linux/kernel/v6.x/linux-6.6.58.tar.xz");
-    curl.follow_location(true);
-    if !(exists(format!("{folder}/kernel.tar.xz")).unwrap()) {
-        let mut out = File::create(format!("{folder}/kernel.tar.xz")).unwrap();
+    if !(exists(format!("{folder}/linux-{kernel_version}.tar.xz")).unwrap()) {
+        let mut curl = easy_curl::new();
+        curl.url(&format!(
+            "https://cdn.kernel.org/pub/linux/kernel/v6.x/linux-{kernel_version}.tar.xz"
+        ))
+        .unwrap();
+        curl.follow_location(true).unwrap();
+        let mut out = File::create(format!("{folder}/linux-{kernel_version}.tar.xz")).unwrap();
         curl.write_function(move |data| {
             out.write_all(data).unwrap();
             Ok(data.len())
-        });
+        })
+        .unwrap();
         curl.perform().unwrap();
     }
 
     // i'm too lazy to do this in rust
-    println!(
-        "{:?}",
-        Command::new("tar")
-            .arg("-xvf")
-            .arg("data/kernel/kernel.tar.xz")
-            .arg("--one-top-level")
-            .current_dir("data/kernel/")
-            .output()
-            .unwrap()
-    );
+    let mut dir = current_dir().unwrap();
+    dir.push(folder);
+    Command::new("tar")
+        .current_dir(dir)
+        .arg("-xf")
+        .arg(&format!("linux-{kernel_version}.tar.xz"))
+        .arg("");
+}
+
+fn grab_datasets() {
+    let kernel_version = "6.6.58";
+
+    create_dir_all("data/kernel").unwrap();
+
+    if !(exists(format!("data/kernel/linux-{kernel_version}")).unwrap()) {
+        println!("Downloading kernel...");
+        grab_kernel("data/kernel".to_string(), kernel_version.to_string());
+        println!("Kernel downloaded");
+    } else {
+        println!("Kernel already downloaded");
+    }
+
+    if !(exists(format!("data/25G-random.bin")).unwrap()) {
+        println!("Generating random 25 GiB file...");
+        large_random_file_generation("data/25G-random.bin".to_string());
+        println!("Random 25 GiB file generated");
+    } else {
+        println!("Random 25 GiB file already generated");
+    }
+
+    if !(exists(format!("data/small-files/random")).unwrap()) {
+        println!("Generating random 1 KiB files...");
+        create_dir_all("data/small-files/random").unwrap();
+        small_random_files_generation("data/small-files/random".to_string());
+        println!("Random 1 KiB files generated...");
+    } else {
+        println!("Random 1 KiB files already generated")
+    }
+
+    if !(exists(format!("data/25G-null.bin")).unwrap()) {
+        println!("Generating empty 25 GiB file...");
+        create_empty_file("data/25G-null.bin".to_string(), 26843545600);
+        println!("Empty 25 GiB file generated...");
+    } else {
+        println!("Empty 25 GiB file already generated");
+    }
+
+    if !(exists("data/small-files/null").unwrap()) {
+        println!("Generating empty 1 KiB files...");
+        create_dir_all("data/small-files/null").unwrap();
+        small_empty_files_generation("data/small-files/null".to_string());
+        println!("Empty 1 KiB files generated...");
+    } else {
+        println!("Empty 1 KiB files already generated")
+    }
+
+    if !(exists("data/small-files/100M-polygon.txt").unwrap()) {
+        println!("*** Get 100M-sided regular polygon data and put it at `./data/small-files/100M-polygon.txt` ***");
+    }
 }
 
 fn main() {
-    large_random_file_generation("data/25G-random.bin".to_string());
-    //single_threaded_large_random_file_generation("data/output".to_string())
-
-    //small_random_files_generation("data/small-files/random".to_string());
-    //grab_kernel("data/kernel".to_string());
+    grab_datasets();
 }