Add the broker-comparison module

Kafka, RabbitMQ and Pulsar measured side by side on ordering, replay and
consumer scaling by driving the three client libraries directly, plus an
operational-footprint measurement of each broker's own distribution.

Nine tests, three brokers started without Docker, and every number in the
documentation regenerated by scripts/run-all.sh.

broker-comparison/scripts/footprint.sh

@@ -0,0 +1,73 @@
+#!/usr/bin/env bash
+# Measures the operational footprint of one broker: how long it takes to accept a connection,
+# how much memory it holds at idle, how many processes and listening ports it opens, and how
+# large its shipped default configuration is.
+#
+# These are proxies, not a benchmark. "Ops burden" is not directly measurable, but the number of
+# moving parts, the memory floor and the size of the configuration surface are, and they are what
+# people are actually asking about when they ask which broker is heavier to run.
+#
+#   scripts/footprint.sh kafka|rabbit|pulsar
+set -euo pipefail
+BROKER="${1:?kafka|rabbit|pulsar}"
+
+extra=""
+port_open() { (echo > "/dev/tcp/127.0.0.1/$1") 2>/dev/null; }
+rss_mb() { ps -eo rss,cmd | grep -F "$1" | grep -v grep | awk '{s+=$1} END {printf "%.0f", s/1024}'; }
+procs()  { ps -eo cmd | grep -F "$1" | grep -cv grep; }
+settings() { grep -cE '^[a-zA-Z]' "$1" 2>/dev/null || echo 0; }
+
+start=$(date +%s%3N)
+case "$BROKER" in
+  kafka)
+    : "${KAFKA_HOME:?set KAFKA_HOME}"
+    # Wipe whatever log.dirs points at, so the format below is not refused with
+    # "Invalid cluster.id ... Expected X, but read Y" from a previous run.
+    KAFKA_LOG_DIRS=$(grep -E '^log.dirs=' "$KAFKA_HOME/config/server.properties" | cut -d= -f2)
+    rm -rf ${KAFKA_LOG_DIRS//,/ }
+    "$KAFKA_HOME/bin/kafka-storage.sh" format --standalone -t "$("$KAFKA_HOME/bin/kafka-storage.sh" random-uuid)" \
+        -c "$KAFKA_HOME/config/server.properties" --ignore-formatted >/dev/null
+    start=$(date +%s%3N)   # after formatting: kafka-storage.sh is two more JVM starts and is a
+                           # one-off, not part of what a restart costs
+    setsid nohup "$KAFKA_HOME/bin/kafka-server-start.sh" "$KAFKA_HOME/config/server.properties" \
+        > /tmp/kafka-start.log 2>&1 < /dev/null &
+    for _ in $(seq 1 90); do port_open 9092 && break; sleep 1; done
+    ready=$(( $(date +%s%3N) - start ))
+    marker="kafka.Kafka"; ports="9092 (broker), 9093 (controller)"; conf="$KAFKA_HOME/config/server.properties"
+    dist="$KAFKA_HOME"
+    ;;
+  rabbit)
+    : "${RABBITMQ_HOME:?set RABBITMQ_HOME}"; : "${ERL_ROOT:?set ERL_ROOT}"
+    scripts/rabbit-broker.sh >/dev/null
+    ready=$(( $(date +%s%3N) - start ))
+    # The broker is an Erlang VM, so the process to measure is beam.smp, not anything named
+    # "rabbitmq". epmd is a second, tiny process and is counted separately below.
+    marker="beam.smp"; ports="5672 (AMQP), 4369 (epmd), 25672 (inter-node)"
+    conf="$RABBITMQ_HOME/etc/rabbitmq/rabbitmq.conf.example"; dist="$RABBITMQ_HOME"
+    extra="epmd processes            : $(procs epmd)
+NOTE: RabbitMQ ships no configuration file at all. rabbitmq.conf.example is
+      entirely commented out and the boot log records \"Config file(s): (none)\".
+      Everything in that file is documentation, not a default."
+    ;;
+  pulsar)
+    : "${PULSAR_HOME:?set PULSAR_HOME}"
+    scripts/pulsar-broker.sh >/dev/null
+    ready=$(( $(date +%s%3N) - start ))
+    marker="PulsarStandaloneStarter"; ports="6650 (binary), 8080 (admin/REST), 2181 (ZooKeeper)"
+    conf="$PULSAR_HOME/conf/standalone.conf"; dist="$PULSAR_HOME"
+    ;;
+  *) echo "unknown broker $BROKER" >&2; exit 2 ;;
+esac
+
+cat <<EOF
+broker                      : $BROKER
+time from launch to first
+  accepted connection       : ${ready} ms
+resident memory at idle     : $(rss_mb "$marker") MB
+server processes            : $(procs "$marker")
+listening ports             : $ports
+unpacked distribution size  : $(du -sm "$dist" | cut -f1) MB
+settings in shipped default
+  configuration file        : $(settings "$conf")   ($(basename "$conf"))
+${extra:-}
+EOF