benchmarks: fix minor issues

betree/haura-benchmarks/haura-plots/haura_plots/__init__.py

@@ -238,10 +238,10 @@ def main():
         print(USAGE_HELP)
         sys.exit(2)
     data = []
-    with open(f"{sys.argv[1]}/betree-metrics.jsonl", 'r', encoding="UTF-8") as metrics:
-        data = util.read_jsonl(metrics)
 
     for path in sys.argv[1:]:
+        with open(f"{path}/betree-metrics.jsonl", 'r', encoding="UTF-8") as metrics:
+            data = util.read_jsonl(metrics)
         # Plot actions
         metrics_plots.plot_throughput(data, path)
         metrics_plots.plot_tier_usage(data, path)

betree/haura-benchmarks/haura-plots/haura_plots/metrics_plots.py

@@ -38,12 +38,12 @@ def plot_throughput(data, path):
             def ms_to_string(time):
                 time: f"{int(time / 1000 / 60)}:{int(time / 1000) % 60:02d}"
 
-            epoch_formatted = list(map(ms_to_string, epoch))
-            axs[tier_id].set_xlabel("runtime (minute:seconds)")
-            axs[tier_id].set_xticks(epoch, epoch_formatted)
-            axs[tier_id].locator_params(tight=True, nbins=10)
-            axs[tier_id].set_ylabel(f"{util.num_to_name(tier_id)}\nMiB/s (I/0)")
-            label=' | '.join(path.split('/')[-2:])
+        epoch_formatted = list(map(ms_to_string, epoch))
+        axs[tier_id].set_xlabel("runtime (minute:seconds)")
+        axs[tier_id].set_xticks(epoch, epoch_formatted)
+        axs[tier_id].locator_params(tight=True, nbins=10)
+        axs[tier_id].set_ylabel(f"{util.num_to_name(tier_id)}\nMiB/s (I/0)")
+        label=' | '.join(path.split('/')[-2:])
     fig.legend(loc="center right",handles=axs[0].get_lines())
     # Epoch in seconds
     fig.suptitle(f"Haura - {label}", y=0.98)  # add title

betree/haura-benchmarks/run.sh

@@ -185,7 +185,7 @@ function ingest() {
   (
     (
       export BETREE__COMPRESSION="None"
-      run "default" ingest_hdd_none ingest "$ZIP_ARCHIVE"
+      run "$RUN_IDENT" ingest_hdd_none ingest "$ZIP_ARCHIVE"
     )
 
     for level in $(seq 1 16); do

betree/haura-benchmarks/src/main.rs

@@ -39,6 +39,8 @@ enum Mode {
         samples: u64,
         min_size: u64,
         max_size: u64,
+        #[structopt(default_value = "0.5")]
+        ratio: f64,
     },
     Zip {
         n_clients: u32,
@@ -106,7 +108,7 @@ fn run_all(mode: Mode) -> Result<(), Box<dyn Error>> {
                 min_size,
                 max_size,
             };
-            scientific_evaluation::run_read(client, config)?;
+            scientific_evaluation::run_read_write(client, config, 1.0)?;
             control.database.write().sync()?;
         }
         Mode::EvaluationRW {
@@ -115,6 +117,7 @@ fn run_all(mode: Mode) -> Result<(), Box<dyn Error>> {
             samples,
             min_size,
             max_size,
+            ratio,
         } => {
             let client = control.client(0, b"scientific_evaluation");
             let config = scientific_evaluation::EvaluationConfig {
@@ -124,7 +127,7 @@ fn run_all(mode: Mode) -> Result<(), Box<dyn Error>> {
                 min_size,
                 max_size,
             };
-            scientific_evaluation::run_read_write(client, config)?;
+            scientific_evaluation::run_read_write(client, config, ratio.clamp(0.0, 1.0))?;
             control.database.write().sync()?;
         }
         Mode::Zip {

betree/haura-benchmarks/src/scientific_evaluation.rs

@@ -43,41 +43,11 @@ fn prepare_store(client: &mut Client, config: &EvaluationConfig) -> Result<(), B
     Ok(())
 }
 
-pub fn run_read(mut client: Client, config: EvaluationConfig) -> Result<(), Box<dyn Error>> {
-    println!("running scientific_evaluation");
-    // Generate positions to read
-    let positions = gen_positions(&mut client, &config);
-    prepare_store(&mut client, &config)?;
-
-    let (obj, _info) = client
-        .object_store
-        .open_object_with_info(OBJ_NAME)?
-        .expect("Object was just created, but can't be opened!");
-
-    let start = std::time::Instant::now();
-    let mut buf = vec![0; config.max_size as usize];
-    let f = std::fs::OpenOptions::new()
-        .write(true)
-        .create(true)
-        .open("evaluation_read.csv")?;
-    let mut w = std::io::BufWriter::new(f);
-    w.write_all(b"offset,size,latency_ns\n")?;
-
-    for (pos, len) in positions.iter().cycle() {
-        // Read data as may be done in some evaluation where only parts of a
-        // database file are read in.
-        let t = std::time::Instant::now();
-        obj.read_at(&mut buf[..*len as usize], *pos).unwrap();
-        w.write_fmt(format_args!("{pos},{len},{}", t.elapsed().as_nanos()))?;
-        if start.elapsed().as_secs() >= config.runtime {
-            break;
-        }
-    }
-    w.flush()?;
-    Ok(())
-}
-
-pub fn run_read_write(mut client: Client, config: EvaluationConfig) -> Result<(), Box<dyn Error>> {
+pub fn run_read_write(
+    mut client: Client,
+    config: EvaluationConfig,
+    rw: f64,
+) -> Result<(), Box<dyn Error>> {
     println!("running scientific_evaluation");
     // Generate positions to read
     let positions = gen_positions(&mut client, &config);
@@ -100,17 +70,17 @@ pub fn run_read_write(mut client: Client, config: EvaluationConfig) -> Result<()
     for (pos, len) in positions.iter().cycle() {
         // Read data as may be done in some evaluation where only parts of a
         // database file are read in.
-        if client.rng.gen_bool(0.5) {
+        if client.rng.gen_bool(rw) {
             let t = std::time::Instant::now();
             obj.read_at(&mut buf[..*len as usize], *pos).unwrap();
-            w.write_fmt(format_args!("{pos},{len},{},r", t.elapsed().as_nanos()))?;
+            w.write_fmt(format_args!("{pos},{len},{},r\n", t.elapsed().as_nanos()))?;
         } else {
             cursor.seek(std::io::SeekFrom::Start(*pos))?;
             let t = std::time::Instant::now();
             with_random_bytes(&mut client.rng, *len, 8 * 1024 * 1024, |b| {
                 cursor.write_all(b)
             })?;
-            w.write_fmt(format_args!("{pos},{len},{},w", t.elapsed().as_nanos()))?;
+            w.write_fmt(format_args!("{pos},{len},{},w\n", t.elapsed().as_nanos()))?;
         }
         if start.elapsed().as_secs() >= config.runtime {
             break;