Batam Dumai Melaka (BDM)

Based on 131 RIPE Atlas measurements from GeoCables monitoring infrastructure, March–April 2026.

The Batam–Dumai–Melaka cable, operated jointly by Moratelindo and Telekom Malaysia since 2009, is the shortest submarine system we currently monitor: 353 kilometres of fibre connecting three landings in the Malacca Strait. Its physics floor — the minimum round-trip time a packet could theoretically achieve end-to-end — is just 3.45 ms.

In one direction, 63 measurements confirm that floor beautifully. From Batam, Indonesia to Melaka, Malaysia, the fastest round-trip we observed is 10.03 ms. That's roughly 2,200 km of optical path at fibre speed — accounting for the transit-network overhead on both ends, it's a clean BDM reading. Packets really are arriving at Melaka through the cable they appear to traverse.

In the reverse direction, the fastest round-trip is 60.63 ms.

That six-fold difference, on the same physical cable, for the same 353 km of geography, is the largest asymmetry in our entire monitoring queue. It's the story of BDM — and the story of how a co-owned regional cable ends up being used very differently by its two co-owners.

What the numbers say

Across 131 measurements spanning seven directions — two domestic (Batam↔Melaka) and five long-haul (eastern European probes reaching Melaka through various transit paths) — BDM's direction-of-travel behaviour splits cleanly in two halves.

The forward direction is what you'd expect from a 353 km cable with a dedicated operator peering: a 10 ms floor that matches the geometry almost to the decimal. The reverse direction shows a 60 ms floor — an entire round-trip that simply cannot be happening through 353 km of fibre. Somewhere on the Malaysian side, the packet is taking a dramatically longer path before it gets to Batam.

Where the reverse direction actually goes

Batam and Melaka are 210 km apart as the crow flies, across one of the busiest waterways in the world. Yet Melaka-to-Batam traffic spends 60 ms at its fastest — enough time, at fibre speed, for a packet to cover roughly 6,000 km. It's clearly not taking the direct BDM path at all.

The most plausible explanation is a Singapore hairpin. Malaysian transit networks — Telekom Malaysia itself, TIME dotCom, Maxis — route most of their international and regional traffic through Singapore's massive peering ecosystem: Equinix SG1, DE-CIX Singapore, Epsilon, and the BGP tables of every major transit provider converge there. For traffic from Melaka toward Batam, the default route is often Melaka → Kuala Lumpur transit → Singapore → Batam, hairpinning south before turning north. The total optical path is easily 500 km or more, and with peering overhead at each hop, a 60 ms minimum is entirely consistent with that geometry.

Moratelindo, on the other hand, has been selling connectivity across BDM since 2009. Indonesian networks routing Batam-bound traffic from Melaka through Moratelindo's own peering get to use the cable directly. Malaysian networks don't — not because they can't, but because their default BGP preferences point at Singapore, and changing that default would require individual peering arrangements that haven't materialised in seventeen years of operation.

Co-ownership isn't routing

BDM is a 50/50 project. Moratelindo and Telekom Malaysia both invested in the cable, both hold capacity rights, and both are listed as owners in TeleGeography's system. What co-ownership does not guarantee is that both partners' commercial networks will actually route their traffic across the cable in both directions. Ownership is a capital structure; routing is a BGP decision, made cable-by-cable, peer-by-peer, and at the operational discretion of each network's routing engineers.

For a cable like MIST — which we published earlier in this same session — the single-operator model (Orient Link) produces much more consistent two-directional behaviour, because the same entity makes BGP decisions on both ends and has uniform commercial incentives. BDM's split ownership exposes the asymmetry directly: each operator optimises for its own traffic and its own peering ecosystem, and the cable ends up as a predominantly one-direction asset in practice.

Other asymmetry cables we've written about

BDM isn't alone in showing this pattern, though its magnitude is unusual:

• EASSy shows a 173 ms reverse asymmetry, driven by operator-preference differences between its African consortium partners.
• PROA, a short Japan-Guam cable, carries a 4.6× reverse asymmetry from NEC-specific transit arrangements.
• Tata TGN Western Europe has bimodal reverse-direction RTT with a 58 ms spread between two stable BGP paths.

BDM's 271 ms symmetry gap is the outlier in this set. It's the combination of co-ownership, a small and commodity-grade physical asset (4 fibre pairs, 2.56 Tbps of total capacity — modest by modern standards), and the gravitational pull that Singapore's peering ecosystem exerts on every piece of Malaysian transit traffic that produces such pronounced extremes.

What HMN Tech supplied

For context, the cable was built by what is now HMN Tech — the submarine division formerly operating as Huawei Marine Networks before the 2020 divestiture. BDM's supplier lineage puts it in the category of cables operated by regional consortia using Chinese-built infrastructure, a category that includes later systems like PEACE. The 2.56 Tbps capacity reflects both its age (the equipment was originally 2009-vintage) and its regional-only ambitions. It was never designed to be a major trans-continental trunk; it was designed to be a short, dedicated Indonesia-Malaysia link, and its forward-direction behaviour shows exactly that — a purpose-built regional asset functioning as intended.

What we're watching

The asymmetry may shift. Malaysia's domestic peering ecosystem has been gradually diversifying over the past few years, with more direct peering agreements emerging between Telekom Malaysia and Indonesian networks. If those agreements start routing Melaka-origin traffic through BDM directly, we'd expect the reverse-direction floor to drop sharply from 60 ms toward 10 ms. That would be the signal that BDM has become a genuinely bidirectional asset instead of a 50/50 cable used 90/10 in practice.

The five international long-haul measurements — from Minsk, Almaty, Tbilisi, Jerusalem, Sevastopol to Melaka — add useful context but don't change the core story. Those all represent traffic reaching Melaka through various long-haul corridors (typically via the SeaMeWe family or the AAE-1 corridor), and their floors of 198–252 ms simply reflect the physics of those 8,000 km journeys, not BDM's own geometry. The core insight is entirely contained in the two short-haul directions, and those two tell a clear story about how commercial routing decisions can make a 353 km cable behave like a 2,000 km cable — in one direction only.