[Chore] Add extra logging for experiments (#19)

* [Chore] Add logging for RocksDB
* [Jupyter] Update jupyter notebooks

.gitignore

@@ -200,3 +200,4 @@ data/*.data
 cmake-build-debug/
 src/bliss/.idea/
 
+db_working_home

script/bench.py

@@ -7,10 +7,7 @@
 from infra.pybliss import BlissArgs, PyBliss
 from infra.util import get_file_params
 
-K_CHOICES = [1, 3, 5, 10, 25, 50]
-L_CHOICES = [1, 3, 5, 10, 25, 50]
-SPECIAL_KL = ((0, 0), (100, 100))
-INDEXES = ["alex", "lipp"]
+INDEXES = ["btree"]
 PRELOAD_FACTOR = 0.4
 WRITE_FACTOR = 0.4
 READ_FACTOR = 0.2
@@ -22,21 +19,22 @@ def main(args):
     bliss = PyBliss(args.bliss, args.smoke_test)
     db = BlissDB(args.result_db)
     files = os.listdir(args.data_folder)
-    exp_pairs = ((file, index) for file in files for index in INDEXES)
+    kwargs = {
+        "preload_factor": PRELOAD_FACTOR,
+        "write_factor": WRITE_FACTOR,
+        "read_factor": READ_FACTOR,
+        "mixed_ratio": MIXED_RATIO,
+        "file_type": "binary",
+        "seed": 0,
+        "use_preload": PRELOAD,
+    }
 
+    exp_pairs = ((file, index) for file in files for index in INDEXES)
     for file, index in exp_pairs:
         _, k_pt, l_pt = get_file_params(file)
         logging.info(f"Running bliss ({index}, {file})")
         bliss_args = BlissArgs(
-            data_file=os.path.join(args.data_folder, file),
-            index_type=index,
-            preload_factor=PRELOAD_FACTOR,
-            write_factor=WRITE_FACTOR,
-            read_factor=READ_FACTOR,
-            mixed_ratio=MIXED_RATIO,
-            file_type="binary",
-            seed=0,
-            use_preload=PRELOAD,
+            data_file=os.path.join(args.data_folder, file), index_type=index, **kwargs
         )
         logging.debug(f"BlissArgs: {bliss_args}")
         stats = bliss.run_single_bliss_bench(bliss_args)

script/data_gen.py

@@ -3,9 +3,12 @@
 import argparse
 from infra.pybods import PyBods, BodsArgs
 
-K_CHOICES = [1, 3, 5, 10, 25, 50, 100]
-L_CHOICES = [1, 3, 5, 10, 25, 50, 100]
-SPECIAL_KL = [(0, 0), (100, 100)]
+# K_CHOICES = [1, 3, 5, 10, 25, 50, 100]
+# L_CHOICES = [1, 3, 5, 10, 25, 50, 100]
+# SPECIAL_KL = [(0, 0), (100, 100)]
+K_CHOICES = [3, 5, 10]
+L_CHOICES = [10]
+SPECIAL_KL = []
 
 NUM_KEYS = 500_000_000
 BINARY_FILE_FORMAT = True
@@ -28,6 +31,7 @@ def main(args: argparse.Namespace):
             k_pt=k_pt,
             l_pt=l_pt,
             binary_file_format=BINARY_FILE_FORMAT,
+            seed=2169,
         )
         results = bods.gen_data(bods_args)
         print(results)

script/infra/pybliss.py

@@ -1,5 +1,6 @@
 from dataclasses import dataclass
 import re
+import os
 import logging
 import subprocess
 import random
@@ -92,6 +93,12 @@ def run_single_bliss_bench(self, args: BlissArgs) -> BlissStats:
         read_time = self.read_time_regex.search(proc_results)
         read_time = int(read_time.group(1)) if read_time else 0
 
+        os.makedirs("./run_logs", exist_ok=True)
+        _, file_name = os.path.split(args.data_file)
+        file_name, _ = os.path.splitext(file_name + f"_{args.index_type}")
+        with open(os.path.join("./run_logs", file_name + ".log"), "w") as fid:
+            fid.write(proc_results)
+
         return BlissStats(
             preload_time=preload_time,
             preload_creation_time=preload_creation,

src/bliss/bench_alex.h

@@ -6,6 +6,7 @@
 #include <vector>
 
 #include "bliss/bliss_index.h"
+#include "spdlog/spdlog.h"
 
 namespace bliss {
 
@@ -28,6 +29,39 @@ class BlissAlexIndex : public BlissIndex<KEY_TYPE, VALUE_TYPE> {
     void put(KEY_TYPE key, VALUE_TYPE value) override {
         _index.insert(key, value);
     }
+
+    void end_routine() override {
+        auto stats = this->_index.stats_;
+        spdlog::info("num_keys = {}", stats.num_keys);
+        spdlog::info("num_model_nodes = {}", stats.num_model_nodes);
+        spdlog::info("num_data_nodes = {}", stats.num_data_nodes);
+        spdlog::info("num_expand_and_scales = {}", stats.num_expand_and_scales);
+        spdlog::info("num_expand_and_retrains = {}",
+                     stats.num_expand_and_retrains);
+        spdlog::info("num_downward_splits = {}", stats.num_downward_splits);
+        spdlog::info("num_sideways_splits = {}", stats.num_sideways_splits);
+        spdlog::info("num_model_node_expansions = {}",
+                     stats.num_model_node_expansions);
+        spdlog::info("num_model_node_splits = {}", stats.num_model_node_splits);
+        spdlog::info("num_downard_split_keys = {}",
+                     stats.num_downward_split_keys);
+        spdlog::info("num_sideways_split_keys = {}",
+                     stats.num_sideways_split_keys);
+        spdlog::info("num_model_node_expansion_pointers = {}",
+                     stats.num_model_node_expansion_pointers);
+        spdlog::info("num_model_node_split_pointers = {}",
+                     stats.num_model_node_split_pointers);
+        spdlog::info("num_node_lookups = {}", stats.num_node_lookups);
+        spdlog::info("num_lookups = {}", stats.num_lookups);
+        spdlog::info("num_inserts = {}", stats.num_inserts);
+        spdlog::info("splitting_time = {}", stats.splitting_time);
+        spdlog::info("cost_computation_time = {}", stats.cost_computation_time);
+
+        spdlog::info("max_fanout = {}",
+                     this->_index.derived_params_.max_fanout);
+        spdlog::info("max_data_node_slots = {}",
+                     this->_index.derived_params_.max_data_node_slots);
+    }
 };
 
 }  // namespace bliss

src/bliss/bench_btree.h

@@ -17,14 +17,16 @@ class BlissBTreeIndex : public BlissIndex<KEY_TYPE, VALUE_TYPE> {
     void bulkload(
         std::vector<std::pair<KEY_TYPE, VALUE_TYPE>> values) override {
         // expects the pairs to be pre-sorted before performing bulk load
-        _index.bulk_load(values.begin(), values.end());
+        this->_index.bulk_load(values.begin(), values.end());
     }
 
-    bool get(KEY_TYPE key) override { return _index.exists(key); }
+    bool get(KEY_TYPE key) override { return this->_index.exists(key); }
 
     void put(KEY_TYPE key, VALUE_TYPE value) override {
-        _index.insert(std::make_pair(key, value));
+        this->_index.insert(std::make_pair(key, value));
     }
+
+    void end_routine() override {}
 };
 
 }  // namespace bliss

src/bliss/bench_lipp.h

@@ -24,6 +24,8 @@ class BlissLippIndex : public BlissIndex<KEY_TYPE, VALUE_TYPE> {
     void put(KEY_TYPE key, VALUE_TYPE value) override {
         _index.insert(key, value);
     }
+
+    void end_routine() override {}
 };
 
 }  // namespace bliss

src/bliss/bliss_index.h

@@ -12,6 +12,7 @@ class BlissIndex {
     virtual void bulkload(std::vector<std::pair<KEY_TYPE, VALUE_TYPE>> values);
     virtual bool get(KEY_TYPE key) = 0;
     virtual void put(KEY_TYPE key, VALUE_TYPE value) = 0;
+    virtual void end_routine() = 0;
 };
 
 }  // namespace bliss

src/bliss_bench.cpp

@@ -270,5 +270,7 @@ int main(int argc, char *argv[]) {
 
     workload_executor(*index, data, config, 0);
 
+    index->end_routine();
+
     return 0;
 }