Micro-mainnet

Logbook.md

@@ -6,6 +6,28 @@
 
 Updated the Leios Slide Deck with reusable slides about this sim
 
+### Micro-mainnet
+
+To accommodate the speed limitations of the Haskell simulation, an even smaller "micro-mainnet" has been created.
+
+- Methodology: [topology-v3.ipynb](data/simulation/pseudo-mainnet/topology-v3.ipynb)
+- Network: [topology-v3.yaml](data/simulation/pseudo-mainnet/topology-v3.yaml)
+- Metrics: [topology-v3.md](data/simulation/pseudo-mainnet/topology-v3.md)
+- Experiment comparing micro- vs mini-mainnet: [analysis.ipynb](analysis/sims/micro-mainnet/analysis.ipynb)
+
+| Metric                       | Value      |
+| ---------------------------: | ---------: |
+| Total nodes                  | 100        |
+| Block producers              | 22         |
+| Relay nodes                  | 78         |
+| Total connections            | 2123       |
+| Network diameter             | 4 hops     |
+| Average connections per node | 21.23      |
+| Average latency              | 97.5 ms    |
+| Maximum latency              | 529.1 ms   |
+| Bidirectional connections    | 389        |
+| Asymmetry ratio              | 63.35%     |
+
 ## 2025-09-11
 
 ### Compendium of data for Delta QSD modeling

analysis/sims/micro-mainnet/combine-results.sh

@@ -0,0 +1,41 @@
+#!/usr/bin/env nix-shell
+#!nix-shell -i bash -p gnused gzip pigz "rWrapper.override { packages = with rPackages; [ data_table R_utils bit64 ggplot2 magrittr stringr ]; }"
+
+set -e
+
+mkdir -p results/$d
+for f in lifecycle resources receipts cpus sizes
+do
+  DIR=$(find experiments -type f -name $f.csv.gz \( -not -empty \) -printf %h\\n -quit)
+  HL=$(sed -n -e '1p' "$DIR/case.csv")
+  HR=$(zcat "$DIR/$f.csv.gz" | sed -n -e '1p')
+  if [[ "$f" == "lifecycle" || "$f" == "resources" || "$f" == "sizes" ]]
+  then
+    FRACT=1.00
+  else
+    FRACT=0.33
+  fi
+  (
+    echo "$HL,$HR"
+    for g in $(find experiments -type f -name $f.csv.gz \( -not -empty \) -printf %h\\n)
+    do
+      if [ ! -e "$g/stderr" ]
+      then
+        echo "Skipping $g because it has no stderr." >> /dev/stderr
+      elif [ -s "$g/stderr" ]
+      then
+        echo "Skipping $g because its stderr is not empty." >> /dev/stderr
+      else
+        BL=$(sed -n -e '2p' "$g/case.csv")
+        zcat "$g/$f.csv.gz" | gawk 'FNR > 1 && rand() <= '"$FRACT"' { print "'"$BL"'" "," $0}'
+      fi
+    done
+  ) | pigz -p 3 -9c > results/$f.csv.gz
+R --vanilla << EOI > /dev/null
+require(data.table)
+sampleSize <- $FRACT
+print(sampleSize)
+$f <- fread("results/$f.csv.gz", stringsAsFactors=TRUE)
+save($f, sampleSize, file="results/$f.Rdata", compression_level=9)
+EOI
+done

analysis/sims/micro-mainnet/experiments.list

@@ -0,0 +1,10 @@
+experiments/topology-v2,0.100/run.sh
+experiments/topology-v3,0.100/run.sh
+experiments/topology-v2,0.150/run.sh
+experiments/topology-v3,0.150/run.sh
+experiments/topology-v2,0.200/run.sh
+experiments/topology-v3,0.200/run.sh
+experiments/topology-v2,0.250/run.sh
+experiments/topology-v2,0.300/run.sh
+experiments/topology-v3,0.250/run.sh
+experiments/topology-v3,0.300/run.sh

analysis/sims/micro-mainnet/experiments/config.yaml

@@ -0,0 +1,110 @@
+
+# Simulation parameters.
+timestamp-resolution-ms: 0.1         
+simulate-transactions: true                    
+cleanup-policies:                        
+- cleanup-expired-vote           
+
+# Protocol parameters for Linear Leios.
+leios-variant: linear-with-tx-references
+linear-eb-propagation-criteria: eb-received
+linear-vote-stage-length-slots: 4
+linear-diffuse-stage-length-slots: 7
+praos-fallback-enabled: true      
+leios-stage-active-voting-slots: 1
+leios-mempool-sampling-strategy: ordered-by-id
+relay-strategy: request-from-first
+treat-blocks-as-full: false                       
+eb-diffusion-strategy: peer-order
+eb-referenced-txs-max-size-bytes: 12000000
+eb-size-bytes-constant: 240
+eb-size-bytes-per-ib: 32
+
+# A conservative upper bound based on empiricial data of empty Praos blocks on
+# Cardano mainnet.
+leios-header-diffusion-time-ms: 1000.0
+
+# Scenario-specific parameters for transaction generation.
+tx-start-time: 60                                               
+tx-stop-time: 960
+tx-generation-distribution:        
+  distribution: constant               
+  value: 10.0                             
+tx-size-bytes-distribution:        
+  distribution: constant      
+  value: 1500              
+tx-conflict-fraction: 0                   
+tx-overcollateralization-factor-distribution:
+  distribution: constant                   
+  value: 0                             
+
+# Based on a linear model fit from `db-analyser` measurements of mainnet blocks
+# for the `Apply` operation, either with or without a term for Plutus execution
+# steps.
+#
+# `Apply CPU [ms]` / `Tx count` ~ (0.6201 ms/tx)
+#
+# `Apply CPU [ms]` ~ 0 + (0.2624 ms/tx) * `Tx count`
+#                      + (0.9487 ms/Gstep) * `Tx exec [Gstep]`
+#
+tx-validation-cpu-time-ms: 0.6201
+
+# Inherited from Cardano mainnet protocol parameters for Praos.
+tx-max-size-bytes: 16384
+rb-generation-probability: 0.05
+rb-head-size-bytes: 1024
+rb-body-max-size-bytes: 90112
+
+# Worst case based on benchmark cluster "Forging: Slot start to announced".
+rb-generation-cpu-time-ms: 71.02
+
+# Based on `apply` statistics for empty blocks using `db-analyser` on blocks
+# from the Cardano mainnet.
+rb-head-validation-cpu-time-ms: 0.4438
+eb-header-validation-cpu-time-ms: 0.4438
+
+# Based on a linear model fit from `db-analyser` measurements of mainnet blocks
+# for the `Reapply` operation, either with or without a term for Plutus
+# execution steps.
+#
+# `Reapply CPU [ms]` ~ (0.3539 ms) + (0.02151 ms/kB) * `Block size [kB]`
+#
+# `Reapply CPU [ms]` ~ (0.3478 ms) + (0.01943 ms/kB) * `Block size [kB]`
+#                                  + (0.02127 ms/Gstep) * `Block exec [Gstep]`
+#
+rb-body-legacy-praos-payload-validation-cpu-time-ms-constant: 0.3539
+rb-body-legacy-praos-payload-validation-cpu-time-ms-per-byte: 0.00002151
+eb-body-validation-cpu-time-ms-constant: 0.3539
+eb-body-validation-cpu-time-ms-per-byte: 0.00002151
+
+# Based on analysis of wFA+LS sortition for realistic stake distribution similar
+# to that of Cardano mainnet, and on the security requirement.
+vote-generation-probability: 600
+vote-threshold: 450
+
+# Based on CDDL for votes, worst case for non-persistent voters.
+vote-bundle-size-bytes-constant: 0
+vote-bundle-size-bytes-per-eb: 164
+
+# Worst case based on benchmarks from cryptography prototype.
+vote-generation-cpu-time-ms-constant: 0.280
+vote-generation-cpu-time-ms-per-tx: 0
+vote-validation-cpu-time-ms: 2.9
+
+# Vote diffusion settings.
+vote-diffusion-strategy: peer-order
+vote-diffusion-max-bodies-to-request: 1
+vote-diffusion-max-headers-to-request: 100
+vote-diffusion-max-window-size: 100
+
+# Worst case based on CDDL for certificates and on empirical study of wFA+LS
+# sortition.
+cert-size-bytes-constant: 8000
+cert-size-bytes-per-node: 0
+
+# Worst case based on benchmarks from cryptography prototype.
+cert-generation-cpu-time-ms-constant: 92.5
+cert-generation-cpu-time-ms-per-node: 0
+cert-validation-cpu-time-ms-constant: 157.2
+cert-validation-cpu-time-ms-per-node: 0
+

analysis/sims/micro-mainnet/experiments/run.sh

@@ -0,0 +1,111 @@
+#!/usr/bin/env nix-shell
+#!nix-shell -i bash -p ansifilter gnugrep gnused gzip pigz bc
+
+set -eo pipefail
+
+cd "$(dirname "${BASH_SOURCE[0]}")"
+
+TX_START=60
+TX_STOP=960
+SIM_STOP=1500
+BW=10
+CPU_COUNT=4
+SIM=Rust
+VARIANT=linear-with-tx-references
+BLOCK_SIZE=12
+TX_SIZE=1500
+LABEL=$(basename "$PWD")
+PROPAGATION=eb-received
+STAGE_LENGTH_DIFF=7
+STAGE_LENGTH_VOTE=4
+NETWORK=$(echo -n "$LABEL" | sed -e 's/^\(.*\),\(.*\)/\1/')
+THROUGHPUT=$(echo -n "$LABEL" | sed -e 's/^\(.*\),\(.*\)/\2/')
+TX_SPACING_HONEST=$(echo "scale=3; $TX_SIZE / $THROUGHPUT / 1000" | bc)
+
+ulimit -S -m 48000000 -v 48000000
+
+if [[ -e sim.log ]]
+then
+  rm sim.log
+fi
+
+mkfifo sim.log
+
+sed -e 's/"bandwidth-bytes-per-second":125000000/"bandwidth-bytes-per-second":'"$((125000 * BW))"'/g' \
+    -e 's/"cpu-core-count":6,/"cpu-core-count":'"$CPU_COUNT"',/g' \
+    "../../../../../data/simulation/pseudo-mainnet/$NETWORK.yaml" \
+  > network.yaml
+
+yaml2json ../config.yaml \
+| jq '. + 
+{
+  "leios-variant": "'"$VARIANT"'"
+, "linear-eb-propagation-criteria": "'"$PROPAGATION"'"
+, "linear-diffuse-stage-length-slots": '"$STAGE_LENGTH_DIFF"'
+, "linear-vote-stage-length-slots": '"$STAGE_LENGTH_VOTE"'
+, "leios-stage-length-slots": '"$STAGE_LENGTH_VOTE"'
+, "eb-referenced-txs-max-size-bytes": ('"$BLOCK_SIZE"' * 1000000)
+, "eb-body-avg-size-bytes": ('"$BLOCK_SIZE"' * 1000000)
+, "tx-size-bytes-distribution": {distribution: "constant", value: '"$TX_SIZE"'}
+, "tx-generation-distribution": {distribution: "constant", value: '"$TX_SPACING_HONEST"'}
+, "tx-start-time": '"$TX_START"'
+, "tx-stop-time": '"$TX_STOP"'
+}
+' > config.yaml
+
+function cleanup() {
+  rm sim.log
+  rm network.yaml
+}
+trap cleanup EXIT
+
+grep -E -v '(Slot|CpuTask|Lottery)' sim.log | pigz -p 3 -9c > sim.log.gz &
+
+case "$SIM" in
+  Rust)
+    ../../sim-cli --parameters config.yaml network.yaml --slots "$SIM_STOP" --conformance-events sim.log > stdout 2> stderr
+    ;;
+  Haskell)
+    ../../ols sim leios --leios-config-file config.yaml --topology-file network.yaml --shared-log-format JSON --conformance-events --output-file sim.log --output-seconds "$SIM_STOP" > stdout 2> stderr
+    ;;
+  *)
+    false
+esac
+
+wait
+
+cat << EOI > case.csv
+Network,Bandwidth,CPU,Diffusion duration,Voting duration,Max EB size,Tx size,Throughput,Tx start [s],Tx stop [s],Sim stop [s]
+$NETWORK,$BW Mb/s,$CPU_COUNT vCPU/node,L_diff = $STAGE_LENGTH_DIFF slots,L_vote = $STAGE_LENGTH_VOTE slots,$BLOCK_SIZE MB/EB,$TX_SIZE B/Tx,$THROUGHPUT TxMB/s,$TX_START,$TX_STOP,$SIM_STOP
+EOI
+
+zcat sim.log.gz \
+| ../../leios-trace-processor \
+  +RTS -N5 -RTS \
+  --trace-file /dev/stdin \
+  --lifecycle-file lifecycle.csv \
+  --cpu-file cpus.csv \
+  --resource-file resources.csv \
+  --receipt-file receipts.csv
+
+(
+  echo 'Kind,Item,Generated [s],Transactions,Endorses'
+  zcat sim.log.gz \
+  | grep -E '(EB|RB)Generated' \
+  | jq -r '
+    .message.type[0:2]
+    + "," + .message.id
+    + "," + (.time_s | tostring)
+    + "," + (.message.transactions | length | tostring)
+    + "," + (if .message.endorsement then .message.endorsement.eb.id else "NA" end)
+  '
+) > sizes.csv
+
+pigz -p 3 -9f {cpus,lifecycle,receipts,resources,sizes}.csv
+
+cat case.csv
+
+if [[ "$SIM" == "Rust" ]]
+then
+  ansifilter stdout | grep -E '^ INFO (praos|leios|network): ' > summary.txt
+fi

analysis/sims/micro-mainnet/experiments/topology-v2,0.100/case.csv

@@ -0,0 +1,2 @@
+Network,Bandwidth,CPU,Diffusion duration,Voting duration,Max EB size,Tx size,Throughput,Tx start [s],Tx stop [s],Sim stop [s]
+topology-v2,10 Mb/s,4 vCPU/node,L_diff = 7 slots,L_vote = 4 slots,12 MB/EB,1500 B/Tx,0.100 TxMB/s,60,960,1500

analysis/sims/micro-mainnet/experiments/topology-v2,0.100/config.yaml

@@ -0,0 +1,67 @@
+{
+  "timestamp-resolution-ms": 0.1,
+  "simulate-transactions": true,
+  "cleanup-policies": [
+    "cleanup-expired-vote"
+  ],
+  "leios-variant": "linear-with-tx-references",
+  "linear-eb-propagation-criteria": "eb-received",
+  "linear-vote-stage-length-slots": 4,
+  "linear-diffuse-stage-length-slots": 7,
+  "praos-fallback-enabled": true,
+  "leios-stage-active-voting-slots": 1,
+  "leios-mempool-sampling-strategy": "ordered-by-id",
+  "relay-strategy": "request-from-first",
+  "treat-blocks-as-full": false,
+  "eb-diffusion-strategy": "peer-order",
+  "eb-referenced-txs-max-size-bytes": 12000000,
+  "eb-size-bytes-constant": 240,
+  "eb-size-bytes-per-ib": 32,
+  "leios-header-diffusion-time-ms": 1000.0,
+  "tx-start-time": 60,
+  "tx-stop-time": 960,
+  "tx-generation-distribution": {
+    "distribution": "constant",
+    "value": 15.000
+  },
+  "tx-size-bytes-distribution": {
+    "distribution": "constant",
+    "value": 1500
+  },
+  "tx-conflict-fraction": 0,
+  "tx-overcollateralization-factor-distribution": {
+    "distribution": "constant",
+    "value": 0
+  },
+  "tx-validation-cpu-time-ms": 0.6201,
+  "tx-max-size-bytes": 16384,
+  "rb-generation-probability": 0.05,
+  "rb-head-size-bytes": 1024,
+  "rb-body-max-size-bytes": 90112,
+  "rb-generation-cpu-time-ms": 71.02,
+  "rb-head-validation-cpu-time-ms": 0.4438,
+  "eb-header-validation-cpu-time-ms": 0.4438,
+  "rb-body-legacy-praos-payload-validation-cpu-time-ms-constant": 0.3539,
+  "rb-body-legacy-praos-payload-validation-cpu-time-ms-per-byte": 0.00002151,
+  "eb-body-validation-cpu-time-ms-constant": 0.3539,
+  "eb-body-validation-cpu-time-ms-per-byte": 0.00002151,
+  "vote-generation-probability": 600,
+  "vote-threshold": 450,
+  "vote-bundle-size-bytes-constant": 0,
+  "vote-bundle-size-bytes-per-eb": 164,
+  "vote-generation-cpu-time-ms-constant": 0.28,
+  "vote-generation-cpu-time-ms-per-tx": 0,
+  "vote-validation-cpu-time-ms": 2.9,
+  "vote-diffusion-strategy": "peer-order",
+  "vote-diffusion-max-bodies-to-request": 1,
+  "vote-diffusion-max-headers-to-request": 100,
+  "vote-diffusion-max-window-size": 100,
+  "cert-size-bytes-constant": 8000,
+  "cert-size-bytes-per-node": 0,
+  "cert-generation-cpu-time-ms-constant": 92.5,
+  "cert-generation-cpu-time-ms-per-node": 0,
+  "cert-validation-cpu-time-ms-constant": 157.2,
+  "cert-validation-cpu-time-ms-per-node": 0,
+  "leios-stage-length-slots": 4,
+  "eb-body-avg-size-bytes": 12000000
+}

analysis/sims/micro-mainnet/experiments/topology-v2,0.100/run.sh

@@ -0,0 +1 @@
+../run.sh