Various improvements to the TPC-C runner (#504)

- Batch inserts - Add Zipfian skew to how keys are selected - Remove use of fetchall() where not needed (does not really impact BRAD though) These changes reduce new order latency from ~95 ms down to ~45 ms. Part of #487.
mitdbg · May 1, 2024 · b9ba0f3 · b9ba0f3
1 parent b4a54a9
commit b9ba0f3
Show file tree

Hide file tree

Showing 11 changed files with 510 additions and 88 deletions.
diff --git a/experiments/17-chbenchmark/common.sh b/experiments/17-chbenchmark/common.sh
@@ -14,13 +14,18 @@ function start_brad() {
 
 function run_tpcc() {
   pushd ../../../workloads/chbenchmark/py-tpcc/
-  RECORD_DETAILED_STATS=1 python3 -m pytpcc.tpcc brad \
-    --no-load \
-    --config $abs_txn_config_file \
-    --warehouses $txn_warehouses \
-    --duration $run_for_s \
-    --clients $t_clients \
-    --scalefactor $txn_scale_factor &
+  local args=(
+    --no-load
+    --config $abs_txn_config_file
+    --warehouses $txn_warehouses
+    --duration $run_for_s
+    --clients $t_clients
+    --scalefactor $txn_scale_factor
+  )
+  if [[ ! -z $txn_zipfian_alpha ]]; then
+    args+=(--zipfian-alpha $txn_zipfian_alpha)
+  fi
+  RECORD_DETAILED_STATS=1 python3 -m pytpcc.tpcc brad "${args[@]}" &
   tpcc_pid=$!
   popd
 }
@@ -91,6 +96,10 @@ function extract_named_arguments() {
     if [[ $phys_arg =~ --txn-config-file=.+ ]]; then
       txn_config_file=${phys_arg:18}
     fi
+
+    if [[ $phys_arg =~ --txn-zipfian-alpha=.+ ]]; then
+      txn_zipfian_alpha=${phys_arg:20}
+    fi
   done
 }
 
diff --git a/experiments/17-chbenchmark/debug/COND b/experiments/17-chbenchmark/debug/COND
@@ -1,3 +1,6 @@
+ZIPFIAN_ALPHA = 5.0
+
+
 run_command(
   name="txn_lat",
   run="./run_tpcc.sh",
@@ -13,6 +16,22 @@ run_command(
   },
 )
 
+run_command(
+  name="txn_lat_zipf",
+  run="./run_tpcc.sh",
+  options={
+    "physical-config-file": "../../../config/physical_config_chbench.yml",
+    "system-config-file": "debug_config.yml",  # Relative to one level up.
+    "txn-config-file": "brad.config",
+    "schema-name": "chbenchmark",
+    "txn-warehouses": 1740,
+    "txn-scale-factor": 1,  # TBD
+    "t-clients": 1,  # TBD
+    "run-for-s": 180,
+    "txn-zipfian-alpha": ZIPFIAN_ALPHA,
+  },
+)
+
 run_command(
   name="aurora_direct",
   run="./run_aurora_direct.sh",
@@ -25,6 +44,19 @@ run_command(
   },
 )
 
+run_command(
+  name="aurora_direct_zipf",
+  run="./run_aurora_direct.sh",
+  options={
+    "txn-config-file": "aurora.config",
+    "txn-warehouses": 1740,
+    "txn-scale-factor": 1,  # TBD
+    "t-clients": 1,  # TBD
+    "run-for-s": 180,
+    "txn-zipfian-alpha": ZIPFIAN_ALPHA,
+  },
+)
+
 run_experiment(
   name="aurora_timing",
   run="./run_aurora_timing.sh",
@@ -36,3 +68,16 @@ run_experiment(
     "run-for-s": 30,
   },
 )
+
+run_experiment(
+  name="aurora_timing_zipf",
+  run="./run_aurora_timing.sh",
+  options={
+    "txn-config-file": "aurora.config",
+    "txn-warehouses": 1740,
+    "txn-scale-factor": 1,  # TBD
+    "t-clients": 1,  # TBD
+    "run-for-s": 30,
+    "txn-zipfian-alpha": ZIPFIAN_ALPHA,
+  },
+)
diff --git a/experiments/17-chbenchmark/debug/run_aurora_direct.sh b/experiments/17-chbenchmark/debug/run_aurora_direct.sh
@@ -9,10 +9,18 @@ extract_named_arguments $@
 abs_txn_config_file=$(realpath $txn_config_file)
 
 cd ../../../workloads/chbenchmark/py-tpcc/
-RECORD_DETAILED_STATS=1 python3 -m pytpcc.tpcc aurora \
-  --no-load \
-  --config $abs_txn_config_file \
-  --warehouses $txn_warehouses \
-  --duration $run_for_s \
-  --clients $t_clients \
+
+args=(
+  --no-load
+  --config $abs_txn_config_file
+  --warehouses $txn_warehouses
+  --duration $run_for_s
+  --clients $t_clients
   --scalefactor $txn_scale_factor
+)
+
+if [[ ! -z $txn_zipfian_alpha ]]; then
+  args+=(--zipfian-alpha $txn_zipfian_alpha)
+fi
+
+RECORD_DETAILED_STATS=1 python3 -m pytpcc.tpcc aurora "${args[@]}"
diff --git a/experiments/17-chbenchmark/debug/run_aurora_timing.sh b/experiments/17-chbenchmark/debug/run_aurora_timing.sh
@@ -9,10 +9,18 @@ extract_named_arguments $@
 abs_txn_config_file=$(realpath $txn_config_file)
 
 cd ../../../workloads/chbenchmark/py-tpcc/
-RECORD_DETAILED_STATS=1 python3 -m pytpcc.tpcc auroratiming \
-  --no-load \
-  --config $abs_txn_config_file \
-  --warehouses $txn_warehouses \
-  --duration $run_for_s \
-  --clients $t_clients \
+
+args=(
+  --no-load
+  --config $abs_txn_config_file
+  --warehouses $txn_warehouses
+  --duration $run_for_s
+  --clients $t_clients
   --scalefactor $txn_scale_factor
+)
+
+if [[ ! -z $txn_zipfian_alpha ]]; then
+  args+=(--zipfian-alpha $txn_zipfian_alpha)
+fi
+
+RECORD_DETAILED_STATS=1 python3 -m pytpcc.tpcc auroratiming "${args[@]}"
diff --git a/tools/calibration/transactions/chbenchmark/COND b/tools/calibration/transactions/chbenchmark/COND
@@ -13,6 +13,8 @@ COND_INSTANCES = {
   instance: instance.replace(".", "_").replace("db.", "") for instance in INSTANCES
 }
 
+ZIPFIAN_ALPHA = 5.0
+
 combine(
   name="all",
   deps=[
@@ -36,6 +38,7 @@ for instance in INSTANCES:
           "txn-warehouses": 1740,
           "txn-config-file": "aurora.config",
           "schema-name": "chbenchmark",
+          "txn-zipfian-alpha": ZIPFIAN_ALPHA,
           "instance": instance,
         },
       )

diff --git a/tools/calibration/transactions/chbenchmark/run_instance.sh b/tools/calibration/transactions/chbenchmark/run_instance.sh
@@ -38,6 +38,10 @@ function extract_named_arguments() {
     if [[ $phys_arg =~ --instance=.+ ]]; then
       instance=${phys_arg:11}
     fi
+
+    if [[ $phys_arg =~ --txn-zipfian-alpha=.+ ]]; then
+      txn_zipfian_alpha=${phys_arg:20}
+    fi
   done
 }
 
@@ -74,7 +78,8 @@ RECORD_DETAILED_STATS=1 python3 -m pytpcc.tpcc aurora \
   --duration $run_for_s \
   --clients $t_clients \
   --scalefactor 1 \
-  --lat-sample-prob 0.25
+  --lat-sample-prob 0.25 \
+  --txn-zipfian-alpha $txn_zipfian_alpha
 popd
 
 >&2 echo "Waiting 10 seconds before retrieving metrics..."