From a4aa1d5f37d6e6df35a844d145abf9f0617552b9 Mon Sep 17 00:00:00 2001
From: Geoffrey Yu <geoffxy@mit.edu>
Date: Tue, 14 Nov 2023 14:33:37 -0500
Subject: [PATCH] Break scale up workload into two parts

---
 experiments/15-e2e-scenarios-v2/scale_up/COND | 42 +++++++++
 .../scale_up/run_overall_warmup.sh            | 37 ++++++++
 .../scale_up/run_workload_ana_up.sh           | 90 +++++++++++++++++++
 .../scale_up/run_workload_txn_up.sh           | 85 ++++++++++++++++++
 4 files changed, 254 insertions(+)
 create mode 100755 experiments/15-e2e-scenarios-v2/scale_up/run_overall_warmup.sh
 create mode 100755 experiments/15-e2e-scenarios-v2/scale_up/run_workload_ana_up.sh
 create mode 100755 experiments/15-e2e-scenarios-v2/scale_up/run_workload_txn_up.sh

diff --git a/experiments/15-e2e-scenarios-v2/scale_up/COND b/experiments/15-e2e-scenarios-v2/scale_up/COND
index 0aac8306..f85b8f16 100644
--- a/experiments/15-e2e-scenarios-v2/scale_up/COND
+++ b/experiments/15-e2e-scenarios-v2/scale_up/COND
@@ -41,3 +41,45 @@ run_command(
     "dataset-type": "100gb",
   },
 )
+
+run_experiment(
+  name="brad_100g_txn_up",
+  run="./run_workload_txn_up.sh",
+  options={
+    # TODO: Ideally, configurations are shared. Only keep AWS secrets separate.
+    "config-file": "config/config_large_100.yml",
+    "planner-config-file": "config/planner.yml",
+    "schema-name": "imdb_extended_100g",
+    "ra-query-bank-file": IMDB_100GB_REGULAR_QUERY_BANK,
+    "num-front-ends": 28,
+    "dataset-type": "100gb",
+  },
+)
+
+run_experiment(
+  name="brad_100g_ana_up",
+  run="./run_workload_ana_up.sh",
+  options={
+    # TODO: Ideally, configurations are shared. Only keep AWS secrets separate.
+    "config-file": "config/config_large_100.yml",
+    "planner-config-file": "config/planner.yml",
+    "schema-name": "imdb_extended_100g",
+    "ra-query-bank-file": IMDB_100GB_REGULAR_QUERY_BANK,
+    "num-front-ends": 28,
+    "dataset-type": "100gb",
+  },
+)
+
+run_command(
+  name="brad_100g_warmup",
+  run="./run_overall_warmup.sh",
+  options={
+    # TODO: Ideally, configurations are shared. Only keep AWS secrets separate.
+    "config-file": "config/config_large_100.yml",
+    "planner-config-file": "config/planner.yml",
+    "schema-name": "imdb_extended_100g",
+    "ra-query-bank-file": IMDB_100GB_REGULAR_QUERY_BANK,
+    "num-front-ends": 28,
+    "dataset-type": "100gb",
+  },
+)
diff --git a/experiments/15-e2e-scenarios-v2/scale_up/run_overall_warmup.sh b/experiments/15-e2e-scenarios-v2/scale_up/run_overall_warmup.sh
new file mode 100755
index 00000000..46d37b61
--- /dev/null
+++ b/experiments/15-e2e-scenarios-v2/scale_up/run_overall_warmup.sh
@@ -0,0 +1,37 @@
+#! /bin/bash
+script_loc=$(cd $(dirname $0) && pwd -P)
+cd $script_loc
+source ../common.sh
+
+initial_queries="99,56,32,92,91,49,30,83,94,38,87,86,76,37,31,46"
+heavier_queries="58,61,62,64,69,73,74,51,57,60"
+
+# Arguments:
+# --config-file
+# --planner-config-file
+# --query-indexes
+extract_named_arguments $@
+
+function inner_cancel_experiment() {
+  cancel_experiment $rana_pid $txn_pid
+}
+
+trap "inner_cancel_experiment" INT
+trap "inner_cancel_experiment" TERM
+
+# Used just to warm up the systems.
+
+export BRAD_IGNORE_BLUEPRINT=1
+start_brad_debug $config_file $planner_config_file
+sleep 30
+
+start_repeating_olap_runner 1 5 5 $initial_queries "ra_8"
+rana_pid=$runner_pid
+sleep 2
+
+log_workload_point "start_txn_2"
+start_txn_runner 2
+txn_pid=$runner_pid
+
+sleep $((5 * 60))
+graceful_shutdown $rana_pid $txn_pid
diff --git a/experiments/15-e2e-scenarios-v2/scale_up/run_workload_ana_up.sh b/experiments/15-e2e-scenarios-v2/scale_up/run_workload_ana_up.sh
new file mode 100755
index 00000000..85eff917
--- /dev/null
+++ b/experiments/15-e2e-scenarios-v2/scale_up/run_workload_ana_up.sh
@@ -0,0 +1,90 @@
+#! /bin/bash
+
+script_loc=$(cd $(dirname $0) && pwd -P)
+cd $script_loc
+source ../common.sh
+
+initial_queries="99,56,32,92,91,49,30,83,94,38,87,86,76,37,31,46"
+heavier_queries="58,61,62,64,69,73,74,51,57,60"
+
+# Arguments:
+# --config-file
+# --planner-config-file
+# --query-indexes
+extract_named_arguments $@
+
+function txn_sweep() {
+  local sweep=$1
+  local gap_minute=$2
+  local keep_last=$3
+
+  for t_clients in $sweep; do
+    start_txn_runner $t_clients  # Implicit: --dataset-type
+    txn_pid=$runner_pid
+
+    sleep $(($gap_minute * 60))
+    if [[ -z $keep_last ]] || [[ $t_clients != $keep_last ]]; then
+      kill -INT $txn_pid
+      wait $txn_pid
+    fi
+  done
+}
+
+function rana_sweep_offset4() {
+  local sweep=$1
+  local gap_minute=$2
+  local keep_last=$3
+  local query_indices=$4
+
+  for ra_clients in $sweep; do
+    start_repeating_olap_runner $ra_clients 15 5 $query_indices "ra_${ra_clients}" 4
+    sweep_rana_pid=$runner_pid
+
+    sleep $(($gap_minute * 60))
+    if [[ -z $keep_last ]] || [[ $ra_clients != $keep_last ]]; then
+      kill -INT $sweep_rana_pid
+      wait $sweep_rana_pid
+    fi
+  done
+}
+
+function inner_cancel_experiment() {
+  if [ ! -z $heavy_rana_pid ]; then
+    cancel_experiment $rana_pid $txn_pid $heavy_rana_pid
+  else
+    cancel_experiment $rana_pid $txn_pid
+  fi
+}
+
+trap "inner_cancel_experiment" INT
+trap "inner_cancel_experiment" TERM
+
+start_brad $config_file $planner_config_file
+log_workload_point "brad_start_initiated"
+sleep 30
+
+# Start with 4 light analytical clients.
+log_workload_point "start_rana_4"
+start_repeating_olap_runner 4 15 5 $initial_queries "ra_4"
+rana_pid=$runner_pid
+log_workload_point "started_rana_4_$rana_pid"
+sleep 2
+
+# Start with 8 transactional clients; hold for 10 minutes to stabilize.
+log_workload_point "start_txn_4"
+start_txn_runner 4
+txn_pid=$runner_pid
+sleep $((10 * 60))
+
+# Scale up to 28 analytical clients in total (24 heavy).
+log_workload_point "start_increase_rana_heavy_4_to_24"
+rana_sweep_offset4 "4 8 12 16 20 24" 3 24 $heavier_queries
+log_workload_point "hold_rana_heavy_24"
+sleep $((30 * 60))  # 18 + 30 mins; 58 cumulative
+
+log_workload_point "experiment_workload_done"
+
+# Shut down everything now.
+>&2 echo "Experiment done. Shutting down runners..."
+graceful_shutdown $rana_pid $heavy_rana_pid $txn_pid
+log_workload_point "shutdown_complete"
diff --git a/experiments/15-e2e-scenarios-v2/scale_up/run_workload_txn_up.sh b/experiments/15-e2e-scenarios-v2/scale_up/run_workload_txn_up.sh
new file mode 100755
index 00000000..e4bf4800
--- /dev/null
+++ b/experiments/15-e2e-scenarios-v2/scale_up/run_workload_txn_up.sh
@@ -0,0 +1,85 @@
+#! /bin/bash
+
+script_loc=$(cd $(dirname $0) && pwd -P)
+cd $script_loc
+source ../common.sh
+
+# Scenario:
+# - Start from 2x t4g.medium Aurora, Redshift off
+# - 4 T clients, increasing to 24 T clients by 4 every minute
+# We expect BRAD to scale up Aurora at this point, but not start Redshift (a replica should be fine for the analytical workload)
+# - Increase the analytical load + add new "heavier" queries - expect that these go to Athena
+# - Increase frequency of queries, expect Redshift to start (go straight to 2x dc2.large to avoid classic resize for practical purposes)
+
+# TODO: This executor file should be adapted to run against the baselines too
+# (TiDB / Serverless Redshift + Aurora)
+
+initial_queries="99,56,32,92,91,49,30,83,94,38,87,86,76,37,31,46"
+heavier_queries="58,61,62,64,69,73,74,51,57,60"
+
+# Arguments:
+# --config-file
+# --planner-config-file
+# --query-indexes
+extract_named_arguments $@
+
+function txn_sweep() {
+  local sweep=$1
+  local gap_minute=$2
+  local keep_last=$3
+
+  for t_clients in $sweep; do
+    start_txn_runner $t_clients  # Implicit: --dataset-type
+    txn_pid=$runner_pid
+
+    sleep $(($gap_minute * 60))
+    if [[ -z $keep_last ]] || [[ $t_clients != $keep_last ]]; then
+      kill -INT $txn_pid
+      wait $txn_pid
+    fi
+  done
+}
+
+function inner_cancel_experiment() {
+  if [ ! -z $heavy_rana_pid ]; then
+    cancel_experiment $rana_pid $txn_pid $heavy_rana_pid
+  else
+    cancel_experiment $rana_pid $txn_pid
+  fi
+}
+
+trap "inner_cancel_experiment" INT
+trap "inner_cancel_experiment" TERM
+
+start_brad $config_file $planner_config_file
+log_workload_point "brad_start_initiated"
+sleep 30
+
+# Start with Aurora 2x db.t4g.medium, Redshift off, Athena unused.
+
+# Start with 4 analytical clients.
+log_workload_point "start_rana_4"
+start_repeating_olap_runner 4 15 5 $initial_queries "ra_4"
+rana_pid=$runner_pid
+log_workload_point "started_rana_4_$rana_pid"
+sleep 2
+
+# Start with 4 transactional clients; hold for 15 minutes to stabilize.
+log_workload_point "start_txn_4"
+start_txn_runner 4
+txn_pid=$runner_pid
+sleep $((10 * 60))  # 10 mins; 10 mins cumulative
+
+# Scale up to 8 transactional clients and hold for 30 minutes.
+log_workload_point "start_increase_txn_4_to_8"
+kill -INT $txn_pid
+wait $txn_pid
+txn_sweep "5 6 7 8" 3 8
+log_workload_point "hold_txn_8_20_min"
+sleep $((35 * 60))  # 47 mins total; 57 mins cumulative
+
+# Shut down everything now.
+log_workload_point "experiment_workload_done"
+>&2 echo "Experiment done. Shutting down runners..."
+graceful_shutdown $rana_pid $heavy_rana_pid $txn_pid
+log_workload_point "shutdown_complete"