Published
Edited
Nov 3, 2020
1 star
md`# Network "Simulator"`
viewof num_nodes = slider({
title: "Number of nodes",
min:0,
max: 100000,
value: 10000,
step: 1000,
description: "Number of nodes in the network",
})
viewof num_links = slider({
min: 0,
title: "Node connections",
max: 1,
step: 0.1,
format: ".0%",
description: "Percentage of nodes a peer is connected to"
})
viewof network_latency = slider({
min: 0,
title: "Network latency (ms)",
max: 2000,
step: 100,
value: 100,
description: "Network latency between nodes"
})
viewof dhtBuckets = slider({
min: 0,
title: "DHT Buckets",
max: 100,
step: 10,
value: 20,
description: "K-buckets in the DHT"
})
viewof ttl = slider({
min: 0,
title: "Bitswap TTL",
max: 20,
step: 1,
value:1,
description: "TTL enabled for Bitswap"
})
// TTL = 0 is vanilla Bitswap
viewof degree = slider({
min: 0,
title: "Bitswap Degree",
max: num_links*num_nodes,
step: 1,
value: 1,
description: "TTL Degree for Bitswap"
})
viewof content_popularity = slider({
min: 1,
title: "Content popularity",
max: num_nodes/10,
step: 10,
value: 1,
description: "Number of nodes storing the content"
})
data = dhtLatency(num_nodes, network_latency, dhtBuckets).concat(ttlLatency(num_nodes, network_latency, ttl, degree, content_popularity))
md`# Lookup latency`
vl.markBar({stroke: "black", fill: "#EEBCDD"})
.height(250)
.width(400)
.data(data)
.encode(
vl.x().fieldN("scheme").sort({field:"scheme", order:"descending"}).scale({paddingInner: .4}),
vl.y().fieldQ("latency"),
vl.tooltip().fieldQ("latency")
)
.render()
prob_data = [{scheme: "DHT", probability:1}, {scheme: "TTL", probability:ttlLatency(num_nodes, network_latency, ttl, degree, content_popularity)[0].probability}]
ttlLatency(num_nodes, network_latency, ttl, degree)
md`# Probability finding content`
vl.markBar({stroke: "black", fill: "#EEBCDD"})
.height(250)
.width(400)
.data(prob_data)
.encode(
vl.x().fieldN("scheme").sort({field:"scheme", order:"descending"}).scale({paddingInner: .4}),
vl.y().fieldQ("probability"),
vl.tooltip().fieldQ("probability")
)
.render()
function connectionLatency(latency){
const rtt = 2*latency
const conn = rtt //Initiate connection
const noise = 1.5*rtt //XX pattern for noise
const multistream = rtt // Multistream negotiation.
const stream_negotiation = rtt // Bitswap or DHT
return conn + noise + multistream
}
function dhtLatency(nodes, latency, kBuckets) {
const conn = connectionLatency(latency)
const reqResp = 2*latency
const worstCase = conn + Math.log2(nodes)*reqResp
const avgCase = conn + Math.log(nodes)/Math.log(kBuckets)*reqResp
//TODO: Take popularity of the content into account in the DHT lookup
return [{scheme: "DHT (avg)", latency: avgCase}, {scheme: "DHT (worst)", latency: worstCase} ]
}
function ttlLatency(nodes, latency, ttl, degree, popularity) {
const conn = connectionLatency(latency)
const reqResp = 4*latency //WANT-HAVE -- WANT-BLOCK
let res = 0
let probability = 0
for (let i = 0; i<=ttl; i++) {
probability += probability_find_ttl(nodes, i, degree, popularity)
res += probability_find_ttl(nodes, i, degree, popularity)*reqResp
}
return [{scheme: "TTL", latency: conn + res, probability: Math.min(1, probability)}]
}
// Probability of finding the content in the current ttl
function probability_find_ttl(nodes, current_ttl, degree, content_popularity){
let subset = nodes/content_popularity
return (Math.pow(degree,current_ttl+1))/(Math.max(1,subset-Math.pow(degree,current_ttl+1)))
}
// Check: https://observablehq.com/@d3/sticky-force-layout?collection=@d3/d3-drag
import {slider} from "@jashkenas/inputs"
import {vl} from "@vega/vega-lite-api"
One platform to build and deploy the best data apps
Experiment and prototype by building visualizations in live JavaScript notebooks. Collaborate with your team and decide which concepts to build out.
Use Observable Framework to build data apps locally. Use data loaders to build in any language or library, including Python, SQL, and R.
Seamlessly deploy to Observable. Test before you ship, use automatic deploy-on-commit, and ensure your projects are always up-to-date.