Nicky_Bao | Rialo

@get_optimum
@RialoHQ

[Nicky_Bao | Rialo @Nicky_Giabao]

How do you measure blockchain propagation latency correctly?

At first glance, the answer seems simple:

Measure how long it takes for a block to reach another node.

In practice, it's much harder than that.

Suppose Protocol A propagates Block #1000 in 150ms, while Protocol B propagates Block #1001 in 900ms.

Can we conclude Protocol A is faster?

Not necessarily.

Different blocks experience different network conditions, different peer topologies, and different levels of congestion. Comparing them introduces measurement bias.

This is exactly the problem Optimum set out to solve when evaluating mump2p on the Ethereum Hoodi testnet.

Instead of measuring different blocks, every block is propagated through both mump2p and Gossipsub simultaneously.

That means:

→ Same block

→ Same origin

→ Same timestamp

→ Same network conditions

The only variable left is the propagation protocol itself.

This "dual path measurement" approach removes one of the biggest sources of benchmarking error and makes the latency comparison statistically meaningful.

That's why the reported 150ms average propagation carries much more weight than a simple benchmark screenshot. @get_optimum

[Nicky_Bao | Rialo @Nicky_Giabao]

Measuring a distributed network requires distributed observations.

A single node can only tell you what happened from its own perspective.

It cannot describe how the entire network behaves.

To solve this, Optimum deployed 30 globally distributed gateways across North America, Europe, and Asia.

Each gateway independently records:

• Slot start time

• Block arrival via Ethereum's native Gossipsub

• Block arrival via mump2p

Those timestamps are collected into Prometheus, where they are aggregated into metrics such as:

→ Average latency

→ p50 / p95 / p99

→ Throughput

→ Message size distribution

→ Peer connectivity

→ Long-term latency trends

One interesting observation is that mump2p averages around 150ms, but periodic spikes to 200–250ms still appear.

That's expected.

Once protocol overhead has been reduced, the remaining delay is largely determined by physical distance and Internet routing.

No networking protocol can eliminate the speed of light.

The goal is to make the protocol efficient enough that geography not software inefficiency becomes the dominant factor. @get_optimum

[Nicky_Bao | Rialo @Nicky_Giabao]

How do you measure blockchain propagation latency correctly?

At first glance, the answer seems simple:

Measure how long it takes for a block to reach another node.

In practice, it's much harder than that.

Suppose Protocol A propagates Block #1000 in 150ms, while Protocol B propagates Block #1001 in 900ms.

Can we conclude Protocol A is faster?

Not necessarily.

Different blocks experience different network conditions, different peer topologies, and different levels of congestion. Comparing them introduces measurement bias.

This is exactly the problem Optimum set out to solve when evaluating mump2p on the Ethereum Hoodi testnet.

Instead of measuring different blocks, every block is propagated through both mump2p and Gossipsub simultaneously.

That means:

→ Same block

→ Same origin

→ Same timestamp

→ Same network conditions

The only variable left is the propagation protocol itself.

This "dual path measurement" approach removes one of the biggest sources of benchmarking error and makes the latency comparison statistically meaningful.

That's why the reported 150ms average propagation carries much more weight than a simple benchmark screenshot. @get_optimum

[Nicky_Bao | Rialo @Nicky_Giabao]

@tung9xnb @get_optimum @blockchainjeff @aqccapital thanks for update

[Nicky_Bao | Rialo @Nicky_Giabao]

@hannie_299 Chị hannie hay thích chạy quá ha

[Nicky_Bao | Rialo @Nicky_Giabao]

@nnthanhthanh96 @ChandlerOtterbe @get_optimum sao mà có thể đỉnh như vậy

[Nicky_Bao | Rialo @Nicky_Giabao]

@jaysonzzzzz_ Jayson on top

[Nicky_Bao | Rialo @Nicky_Giabao]

@DatQuoc62246947 great post

[Nicky_Bao | Rialo @Nicky_Giabao]

@DatQuoc62246947 content chất lượng quá

[Nicky_Bao | Rialo @Nicky_Giabao]

@PhucvinhN1965 GRialo a

[Nicky_Bao | Rialo @Nicky_Giabao]

@thoai6sixxx @RialoHQ GRialo a

[Nicky_Bao | Rialo @Nicky_Giabao]

@VngQuangVinh10 congrat to all

[Nicky_Bao | Rialo @Nicky_Giabao]

@BaeSynA @get_optimum nice art Syn

[Nicky_Bao | Rialo @Nicky_Giabao]

@novanguyen2802 nice article a

[Nicky_Bao | Rialo @Nicky_Giabao]

@Martin_2368 @get_optimum congrat all member

[Nicky_Bao | Rialo @Nicky_Giabao]

@hannie_299 chời ơi sao mà vẽ xịn vậy hả

[Nicky_Bao | Rialo @Nicky_Giabao]

@An03894438 @get_optimum @blockchainjeff Richman hả a

[Nicky_Bao | Rialo @Nicky_Giabao]

@An03894438 GDoma

[Nicky_Bao | Rialo @Nicky_Giabao]

@An03894438 nice article síp

[Nicky_Bao | Rialo @Nicky_Giabao]

@An03894438 @khant156x Congrat to all

[Nicky_Bao | Rialo @Nicky_Giabao]

@novanguyen2802 @RialoHQ GRialo a

[Nicky_Bao | Rialo @Nicky_Giabao]

@Richardx1202 ảnh ngầu qua a nè