first commit

exercises/basic/Makefile

@@ -1,5 +1,5 @@
 # SPDX-License-Identifier: Apache-2.0
 BMV2_SWITCH_EXE = simple_switch_grpc
-TOPO = pod-topo/topology.json
+# TOPO = triangle-topo/topology.json
 
 include ../../utils/Makefile

exercises/basic/analysis/flow_completion.statistics

@@ -0,0 +1,89 @@
+all_99.9th_fct_ms=14578.498400000004
+all_99.9th_fct_ns=14578498400.000004
+all_99th_fct_ms=13664.98400000001
+all_99th_fct_ns=13664984000.00001
+all_flows_completed_fraction=1.0
+all_mean_fct_ms=2532.5430555555554
+all_mean_fct_ns=2532543055.5555553
+all_median_fct_ms=692.7
+all_median_fct_ns=692700000.0
+all_num_finished_flows=72
+all_num_flows=72
+all_num_unfinished_flows=0
+general_flow_size_bytes_mean=1573375.0
+general_flow_size_bytes_std=3608652.9321029196
+general_num_flows=72
+general_num_unique_sources=72
+general_num_unique_targets=72
+geq_100KB_99.9th_fct_ms=14619.956800000007
+geq_100KB_99.9th_fct_ns=14619956800.000008
+geq_100KB_99th_fct_ms=14079.567999999997
+geq_100KB_99th_fct_ns=14079567999.999998
+geq_100KB_flows_completed_fraction=1.0
+geq_100KB_mean_fct_ms=4162.816279069768
+geq_100KB_mean_fct_ns=4162816279.0697675
+geq_100KB_median_fct_ms=2875.9
+geq_100KB_median_fct_ns=2875900000.0
+geq_100KB_num_finished_flows=43
+geq_100KB_num_flows=43
+geq_100KB_num_unfinished_flows=0
+geq_10MB_99.9th_fct_ms=14678.5704
+geq_10MB_99.9th_fct_ns=14678570400.0
+geq_10MB_99th_fct_ms=14665.704
+geq_10MB_99th_fct_ns=14665704000.0
+geq_10MB_flows_completed_fraction=1.0
+geq_10MB_mean_fct_ms=13965.2
+geq_10MB_mean_fct_ns=13965200000.0
+geq_10MB_median_fct_ms=13965.2
+geq_10MB_median_fct_ns=13965200000.0
+geq_10MB_num_finished_flows=2
+geq_10MB_num_flows=2
+geq_10MB_num_unfinished_flows=0
+geq_1MB_99.9th_fct_ms=14651.408000000005
+geq_1MB_99.9th_fct_ns=14651408000.000006
+geq_1MB_99th_fct_ms=14394.080000000002
+geq_1MB_99th_fct_ns=14394080000.000002
+geq_1MB_flows_completed_fraction=1.0
+geq_1MB_mean_fct_ms=7226.92380952381
+geq_1MB_mean_fct_ns=7226923809.523809
+geq_1MB_median_fct_ms=6346.5
+geq_1MB_median_fct_ns=6346500000.0
+geq_1MB_num_finished_flows=21
+geq_1MB_num_flows=21
+geq_1MB_num_unfinished_flows=0
+geq_2.4349MB_99.9th_fct_ms=14661.415200000001
+geq_2.4349MB_99.9th_fct_ns=14661415200.000002
+geq_2.4349MB_99th_fct_ms=14494.151999999998
+geq_2.4349MB_99th_fct_ns=14494151999.999998
+geq_2.4349MB_flows_completed_fraction=1.0
+geq_2.4349MB_mean_fct_ms=9018.371428571429
+geq_2.4349MB_mean_fct_ns=9018371428.571428
+geq_2.4349MB_median_fct_ms=9785.05
+geq_2.4349MB_median_fct_ns=9785050000.0
+geq_2.4349MB_num_finished_flows=14
+geq_2.4349MB_num_flows=14
+geq_2.4349MB_num_unfinished_flows=0
+less_100KB_99.9th_fct_ms=460.2460000000001
+less_100KB_99.9th_fct_ns=460246000.0000001
+less_100KB_99th_fct_ms=439.95999999999987
+less_100KB_99th_fct_ns=439959999.9999999
+less_100KB_flows_completed_fraction=1.0
+less_100KB_mean_fct_ms=115.24137931034483
+less_100KB_mean_fct_ns=115241379.31034483
+less_100KB_median_fct_ms=14.8
+less_100KB_median_fct_ns=14800000.0
+less_100KB_num_finished_flows=29
+less_100KB_num_flows=29
+less_100KB_num_unfinished_flows=0
+less_2.4349MB_99.9th_fct_ms=4257.393900000002
+less_2.4349MB_99.9th_fct_ns=4257393900.000002
+less_2.4349MB_99th_fct_ms=4063.839
+less_2.4349MB_99th_fct_ns=4063839000.0
+less_2.4349MB_flows_completed_fraction=1.0
+less_2.4349MB_mean_fct_ms=966.998275862069
+less_2.4349MB_mean_fct_ns=966998275.862069
+less_2.4349MB_median_fct_ms=422.25
+less_2.4349MB_median_fct_ns=422250000.0
+less_2.4349MB_num_finished_flows=58
+less_2.4349MB_num_flows=58
+less_2.4349MB_num_unfinished_flows=0

exercises/basic/analysis/flow_completion_ct.statistics

@@ -0,0 +1,89 @@
+all_99.9th_fct_ms=14276.2466
+all_99.9th_fct_ns=14276246600.0
+all_99th_fct_ms=14036.366000000002
+all_99th_fct_ns=14036366000.000002
+all_flows_completed_fraction=1.0
+all_mean_fct_ms=2498.5416666666665
+all_mean_fct_ns=2498541666.6666665
+all_median_fct_ms=625.15
+all_median_fct_ns=625150000.0
+all_num_finished_flows=72
+all_num_flows=72
+all_num_unfinished_flows=0
+general_flow_size_bytes_mean=1573375.0
+general_flow_size_bytes_std=3608652.9321029196
+general_num_flows=72
+general_num_unique_sources=72
+general_num_unique_targets=72
+geq_100KB_99.9th_fct_ms=14287.133200000002
+geq_100KB_99.9th_fct_ns=14287133200.000002
+geq_100KB_99th_fct_ms=14145.232
+geq_100KB_99th_fct_ns=14145232000.0
+geq_100KB_flows_completed_fraction=1.0
+geq_100KB_mean_fct_ms=4109.076744186046
+geq_100KB_mean_fct_ns=4109076744.1860466
+geq_100KB_median_fct_ms=2603.9
+geq_100KB_median_fct_ns=2603900000.0
+geq_100KB_num_finished_flows=43
+geq_100KB_num_flows=43
+geq_100KB_num_unfinished_flows=0
+geq_10MB_99.9th_fct_ms=14302.5246
+geq_10MB_99.9th_fct_ns=14302524600.0
+geq_10MB_99th_fct_ms=14299.146
+geq_10MB_99th_fct_ns=14299146000.0
+geq_10MB_flows_completed_fraction=1.0
+geq_10MB_mean_fct_ms=14115.2
+geq_10MB_mean_fct_ns=14115200000.0
+geq_10MB_median_fct_ms=14115.2
+geq_10MB_median_fct_ns=14115200000.0
+geq_10MB_num_finished_flows=2
+geq_10MB_num_flows=2
+geq_10MB_num_unfinished_flows=0
+geq_1MB_99.9th_fct_ms=14295.392000000002
+geq_1MB_99.9th_fct_ns=14295392000.000002
+geq_1MB_99th_fct_ms=14227.82
+geq_1MB_99th_fct_ns=14227820000.0
+geq_1MB_flows_completed_fraction=1.0
+geq_1MB_mean_fct_ms=7281.7
+geq_1MB_mean_fct_ns=7281700000.0
+geq_1MB_median_fct_ms=6908.6
+geq_1MB_median_fct_ns=6908600000.0
+geq_1MB_num_finished_flows=21
+geq_1MB_num_flows=21
+geq_1MB_num_unfinished_flows=0
+geq_2.4349MB_99.9th_fct_ms=14298.0198
+geq_2.4349MB_99.9th_fct_ns=14298019800.0
+geq_2.4349MB_99th_fct_ms=14254.098
+geq_2.4349MB_99th_fct_ns=14254098000.0
+geq_2.4349MB_flows_completed_fraction=1.0
+geq_2.4349MB_mean_fct_ms=9170.507142857143
+geq_2.4349MB_mean_fct_ns=9170507142.857143
+geq_2.4349MB_median_fct_ms=9693.1
+geq_2.4349MB_median_fct_ns=9693100000.0
+geq_2.4349MB_num_finished_flows=14
+geq_2.4349MB_num_flows=14
+geq_2.4349MB_num_unfinished_flows=0
+less_100KB_99.9th_fct_ms=349.8824000000001
+less_100KB_99.9th_fct_ns=349882400.00000006
+less_100KB_99th_fct_ms=342.52399999999994
+less_100KB_99th_fct_ns=342523999.99999994
+less_100KB_flows_completed_fraction=1.0
+less_100KB_mean_fct_ms=110.50689655172414
+less_100KB_mean_fct_ns=110506896.55172414
+less_100KB_median_fct_ms=28.8
+less_100KB_median_fct_ns=28800000.0
+less_100KB_num_finished_flows=29
+less_100KB_num_flows=29
+less_100KB_num_unfinished_flows=0
+less_2.4349MB_99.9th_fct_ms=4804.555900000004
+less_2.4349MB_99.9th_fct_ns=4804555900.000004
+less_2.4349MB_99th_fct_ms=4388.359
+less_2.4349MB_99th_fct_ns=4388359000.0
+less_2.4349MB_flows_completed_fraction=1.0
+less_2.4349MB_mean_fct_ms=888.0672413793104
+less_2.4349MB_mean_fct_ns=888067241.3793104
+less_2.4349MB_median_fct_ms=355.9
+less_2.4349MB_median_fct_ns=355900000.0
+less_2.4349MB_num_finished_flows=58
+less_2.4349MB_num_flows=58
+less_2.4349MB_num_unfinished_flows=0

exercises/basic/analysis/throughput.statistics

@@ -0,0 +1,60 @@
+all_num_flows=72
+all_throughput_0.1th_Mbps=1.53414
+all_throughput_1th_Mbps=1.7514
+all_throughput_99.9th_Mbps=36.16870000000002
+all_throughput_99th_Mbps=29.587000000000064
+all_throughput_mean_Mbps=8.248888888888887
+all_throughput_median_Mbps=6.0600000000000005
+all_throughput_std_Mbps=6.908011965897834
+general_flow_size_bytes_mean=1574295.608888889
+general_num_flows=72
+general_num_unique_sources=72
+general_num_unique_targets=72
+general_throughput_mbps_mean=8.248888888888887
+geq_100KB_num_flows=43
+geq_100KB_throughput_0.1th_Mbps=2.15588
+geq_100KB_throughput_1th_Mbps=2.2088
+geq_100KB_throughput_99.9th_Mbps=26.553800000000006
+geq_100KB_throughput_99th_Mbps=26.137999999999998
+geq_100KB_throughput_mean_Mbps=7.761860465116279
+geq_100KB_throughput_median_Mbps=5.83
+geq_100KB_throughput_std_Mbps=5.875536692285266
+geq_10MB_num_flows=2
+geq_10MB_throughput_1th_Mbps=25.511
+geq_10MB_throughput_99.9th_Mbps=26.5989
+geq_10MB_throughput_99th_Mbps=26.589000000000002
+geq_10MB_throughput_mean_Mbps=26.05
+geq_10MB_throughput_median_Mbps=26.05
+geq_10MB_throughput_std_Mbps=0.5500000000000007
+geq_1MB_num_flows=21
+geq_1MB_throughput_0.1th_Mbps=4.469
+geq_1MB_throughput_1th_Mbps=4.55
+geq_1MB_throughput_99.9th_Mbps=26.578000000000007
+geq_1MB_throughput_99th_Mbps=26.380000000000003
+geq_1MB_throughput_mean_Mbps=11.404285714285717
+geq_1MB_throughput_median_Mbps=8.86
+geq_1MB_throughput_std_Mbps=6.559180096430877
+geq_2.4349MB_num_flows=14
+geq_2.4349MB_throughput_0.1th_Mbps=4.48249
+geq_2.4349MB_throughput_1th_Mbps=4.6849
+geq_2.4349MB_throughput_99.9th_Mbps=26.585700000000003
+geq_2.4349MB_throughput_99th_Mbps=26.457
+geq_2.4349MB_throughput_mean_Mbps=13.345714285714289
+geq_2.4349MB_throughput_median_Mbps=12.25
+geq_2.4349MB_throughput_std_Mbps=7.180166844744442
+less_100KB_num_flows=29
+less_100KB_throughput_0.1th_Mbps=1.51952
+less_100KB_throughput_1th_Mbps=1.6052
+less_100KB_throughput_99.9th_Mbps=36.564000000000014
+less_100KB_throughput_99th_Mbps=33.539999999999985
+less_100KB_throughput_mean_Mbps=8.971034482758622
+less_100KB_throughput_median_Mbps=6.3
+less_100KB_throughput_std_Mbps=8.149718567100884
+less_2.4349MB_num_flows=58
+less_2.4349MB_throughput_0.1th_Mbps=1.52938
+less_2.4349MB_throughput_1th_Mbps=1.7038
+less_2.4349MB_throughput_99.9th_Mbps=36.21600000000006
+less_2.4349MB_throughput_99th_Mbps=30.059999999999995
+less_2.4349MB_throughput_mean_Mbps=7.018620689655173
+less_2.4349MB_throughput_median_Mbps=5.205
+less_2.4349MB_throughput_std_Mbps=6.245888150388073

exercises/basic/analyze_fct.py

@@ -0,0 +1,150 @@
+#!/usr/bin/env python3
+import numpy as np
+import csv
+import sys
+import os
+import pandas as pd
+import matplotlib.pyplot as plt
+
+##################################
+# Setup
+#
+
+print("Flow Completion Time Analysis Tool v0.01")
+
+# Create analysis folder if it doesn't exist
+analysis_folder_path = 'analysis'
+if not os.path.exists(analysis_folder_path):
+    os.makedirs(analysis_folder_path)
+
+##################################
+# Analyze flow completion
+#
+def analyze_flow_completion(csv_file_path):
+    print(f"Analyzing flow completion times from: {csv_file_path}")
+
+    # Read the CSV file using pandas
+    df = pd.read_csv(csv_file_path)
+
+    # Extract relevant columns
+    flow_ids = df['flow_id'].values
+    source_ids = df['sender'].values
+    target_ids = df['receiver'].values
+    flow_sizes = df['flow_size'].values
+    durations = df['duration'].values  # These are already in seconds
+    progress = df['progress'].values
+
+    # Determine completion status (considering flows with ≥99.9% progress as completed)
+    completed = progress >= 99.9
+
+    print("Calculating statistics...")
+
+    statistics = {
+        'general_num_flows': len(flow_ids),
+        'general_num_unique_sources': len(set(source_ids)),
+        'general_num_unique_targets': len(set(target_ids)),
+        'general_flow_size_bytes_mean': np.mean(flow_sizes),
+        'general_flow_size_bytes_std': np.std(flow_sizes)
+    }
+
+    # Define flow size ranges
+    range_low =                     [-1,            -1,            -1,              100000,     2434900,            1000000,    10000000]
+    range_high =                    [-1,            100000,        2434900,         -1,         -1,                 -1,         -1]
+    range_name =                    ["all",         "less_100KB",  "less_2.4349MB", "geq_100KB", "geq_2.4349MB",    "geq_1MB",  "geq_10MB"]
+    range_completed_duration =      [[],            [],            [],              [],         [],                 [],         []]
+    range_num_finished_flows =      [0,             0,             0,               0,          0,                  0,          0]
+    range_num_unfinished_flows =    [0,             0,             0,               0,          0,                  0,          0]
+    range_low_eq =                  [0,             0,             0,               1,          1,                  1,          1]
+    range_high_eq =                 [0,             0,             0,               1,          1,                  1,          1]
+
+    # Go over all flows
+    for i in range(0, len(flow_ids)):
+        # Range-specific
+        for j in range(0, len(range_name)):
+            if (
+                    (range_low[j] == -1 or (range_low_eq[j] == 0 and flow_sizes[i] > range_low[j]) or (range_low_eq[j] == 1 and flow_sizes[i] >= range_low[j])) and
+                    (range_high[j] == -1 or (range_high_eq[j] == 0 and flow_sizes[i] < range_high[j]) or (range_high_eq[j] == 1 and flow_sizes[i] <= range_high[j]))
+            ):
+                if completed[i]:
+                    range_num_finished_flows[j] += 1
+                    range_completed_duration[j].append(durations[i] * 1000000000)  # Convert seconds to nanoseconds
+                else:
+                    range_num_unfinished_flows[j] += 1
+
+    # Ranges statistics
+    for j in range(0, len(range_name)):
+        # Number of finished flows
+        statistics[range_name[j] + '_num_flows'] = range_num_finished_flows[j] + range_num_unfinished_flows[j]
+        print(range_name[j] + '_num_flows', range_num_finished_flows[j] + range_num_unfinished_flows[j])
+
+        statistics[range_name[j] + '_num_finished_flows'] = range_num_finished_flows[j]
+        print(range_name[j] + '_num_finished_flows', range_num_finished_flows[j])
+
+        statistics[range_name[j] + '_num_unfinished_flows'] = range_num_unfinished_flows[j]
+        print(range_name[j] + '_num_unfinished_flows', range_num_unfinished_flows[j])
+
+        total = (range_num_finished_flows[j] + range_num_unfinished_flows[j])
+        if range_num_finished_flows[j] != 0:
+            statistics[range_name[j] + '_flows_completed_fraction'] = float(range_num_finished_flows[j]) / float(total)
+
+            # Duration is stored in nanoseconds in the statistics
+            statistics[range_name[j] + '_mean_fct_ns'] = np.mean(range_completed_duration[j])
+            print(range_name[j] + '_mean_fct_ns', np.mean(range_completed_duration[j]))
+
+            statistics[range_name[j] + '_median_fct_ns'] = np.median(range_completed_duration[j])
+            statistics[range_name[j] + '_99th_fct_ns'] = np.percentile(range_completed_duration[j], 99)
+            statistics[range_name[j] + '_99.9th_fct_ns'] = np.percentile(range_completed_duration[j], 99.9)
+
+            # Convert to milliseconds for display purposes
+            statistics[range_name[j] + '_mean_fct_ms'] = statistics[range_name[j] + '_mean_fct_ns'] / 1000000
+            statistics[range_name[j] + '_median_fct_ms'] = statistics[range_name[j] + '_median_fct_ns'] / 1000000
+            statistics[range_name[j] + '_99th_fct_ms'] = statistics[range_name[j] + '_99th_fct_ns'] / 1000000
+            statistics[range_name[j] + '_99.9th_fct_ms'] = statistics[range_name[j] + '_99.9th_fct_ns'] / 1000000
+        else:
+            statistics[range_name[j] + '_flows_completed_fraction'] = 0
+
+    # Add the original duration values for comparing with other analysis
+    print("\nOriginal duration values (seconds) for comparison:")
+    print(f"All flows - Mean duration: {np.mean(durations):.4f} seconds")
+    print(f"Small flows (<100KB) - Mean duration: {np.mean(durations[flow_sizes <= 100000]):.4f} seconds")
+    print(f"Medium flows (>100KB, <2.4MB) - Mean duration: {np.mean(durations[(flow_sizes > 100000) & (flow_sizes < 2434900)]):.4f} seconds")
+    print(f"Large flows (>1MB) - Mean duration: {np.mean(durations[flow_sizes > 1000000]):.4f} seconds")
+
+    # Create CDF plot for flow completion times
+    plt.figure(figsize=(10, 6))
+    for j in range(len(range_name)):
+        if range_num_finished_flows[j] > 0:
+            sorted_fct = np.sort(range_completed_duration[j]) / 1000000  # Convert ns to ms for readability
+            cdf = np.arange(1, len(sorted_fct)+1) / len(sorted_fct)
+            plt.plot(sorted_fct, cdf, label=range_name[j])
+
+    plt.xlabel('Flow Completion Time (ms)')
+    plt.ylabel('CDF')
+    plt.title('CDF of Flow Completion Times by Flow Size Category')
+    plt.grid(True, linestyle='--', linewidth=0.5)
+    plt.legend()
+    plt.savefig(f'{analysis_folder_path}/fct_cdf.png')
+
+    # Print raw results
+    print('Writing to result file flow_completion.statistics...')
+    with open(analysis_folder_path + '/flow_completion.statistics', 'w+') as outfile:
+        for key, value in sorted(statistics.items()):
+            outfile.write(str(key) + "=" + str(value) + "\n")
+
+    # Print summary statistics for comparison
+    print("\nSummary of Flow Completion Times:")
+    print("Flow Size Category      | Mean (sec)  | Median (sec) | 99th (sec)")
+    print("------------------------|-------------|--------------|------------")
+    for j in range(len(range_name)):
+        if range_num_finished_flows[j] > 0:
+            mean_sec = statistics[range_name[j] + '_mean_fct_ns'] / 1e9
+            median_sec = statistics[range_name[j] + '_median_fct_ns'] / 1e9
+            p99_sec = statistics[range_name[j] + '_99th_fct_ns'] / 1e9
+            print(f"{range_name[j].ljust(24)} | {mean_sec:11.4f} | {median_sec:12.4f} | {p99_sec:10.4f}")
+
+if __name__ == "__main__":
+    # Path to the CSV file
+    csv_file_path = "flow_analysis.csv"
+
+    # Analyze flow completion
+    analyze_flow_completion(csv_file_path)