Dumai-Melaka Cable System (DMCS)

Based on 193 RIPE Atlas measurements from GeoCables monitoring infrastructure, March–April 2026.

The Dumai-Melaka Cable System — DMCS — is a 159-kilometre submarine cable that crosses the Strait of Malacca at its central, busiest section. The Indonesian landing is at Dumai, an oil-port city on the eastern coast of Sumatra. The Malaysian landing is at Melaka — better known by its English transliteration Malacca — the historic UNESCO-listed trading port on Peninsular Malaysia's southwestern coast. The cable was placed in service in 2005, has been operational for twenty-one years, and is owned jointly by Telekom Malaysia and Telkom Indonesia: the two countries' national incumbent telecom operators, in a rare bilateral state-level cable agreement.

The strait that DMCS crosses is one of the most heavily trafficked stretches of water on the planet. Roughly ninety thousand commercial ships transit the Strait of Malacca each year — about a quarter of global maritime trade by volume. Underneath that traffic, on the seabed, the strait carries fewer cables than its strategic importance might suggest, and DMCS is one of the few that connects its two shores rather than running parallel to it. The narrowest point of the strait is about 25 nautical miles wide; DMCS's 159-kilometre route includes the shore approaches and the necessary detours around the strait's busiest shipping channels.

Light in optical fibre takes about 1.56 ms to round-trip 159 kilometres. The minimum we have measured between probes on the two sides is 60.87 ms — roughly 39× the physics floor. For a cable this short, sitting between two state operators with explicit cooperation, that multiplier is already more than the link itself can be doing. But the floor multiplier is not the most striking number on DMCS. The standard deviation is.

Variance, not asymmetry, defines the story

Across 90 RIPE Atlas measurements from Melaka (MY) to Dumai (ID), the average round-trip is 166.85 ms. The maximum we have observed in the same measurement set is 323.80 ms. The standard deviation across the 90 samples is 99.15 ms.

To put that in perspective: the same probe set, measuring from Minsk to Dumai over a path that crosses ten thousand kilometres and four to six independent cable systems, produces a standard deviation of 0.27 ms. The cable that should be the cleanest measurement on the system — 159 kilometres of glass between two state-owned terminal points — has variance several hundred times larger than a path that includes the entire intercontinental backbone between Belarus and Sumatra. Sometimes a packet on DMCS comes back in 60 ms. Sometimes it comes back in 320 ms. Both happen on the same cable, between the same probes, within the same measurement window, with no announced route flap on either operator's side.

The variance is not consistent with a stable single path. It is consistent with multiple alternative paths through the broader Singapore–Kuala Lumpur peering fabric, with packets being load-balanced or dynamically rerouted between them depending on which is least loaded at the moment of measurement. Some of those paths are short; some take packets through Singapore IXPs and back; some appear to take much longer detours through carrier hubs we have not been able to attribute confidently.

State cooperation, market routing

DMCS is, on paper, a clear case of bilateral state cooperation. Telekom Malaysia and Telkom Indonesia jointly funded and operate the cable. The terminal stations are inside national infrastructure on both sides. By any conventional reading of submarine cable economics, traffic exchanged between Indonesian and Malaysian destinations should preferentially traverse DMCS, because the per-bit cost on a jointly owned bilateral cable is lower than the cost of paying transit through an external IXP.

The measurements suggest that the conventional reading does not survive contact with day-to-day BGP. Most ID-MY traffic, including traffic between the carriers' own subscriber networks, evidently still routes through Singapore — the regional carrier hub that sits less than 200 km north of the cable's western end. Singapore hosts the cheapest, deepest peering ecosystem in Southeast Asia, and BGP path selection responds to that price-and-density gravity even when an alternative direct path exists between the source and destination ASNs.

What DMCS measures, then, is not the link. It is the persistent gap between bilateral cable diplomacy and the actual path that packets take. Two state operators built a cable. Their networks route around it.

Geography and traffic don't agree

The most direct comparison in our monitoring set is BRCS (Batam-Rengit Cable System), which connects two points sixty-four kilometres apart across a different stretch of the same Singapore-Strait region. BRCS measures at 80× its physics floor; DMCS at 39×. BRCS shows extreme directional asymmetry — forward and reverse take different oceans' worth of detour. DMCS shows extreme variance instead — the path is not consistent enough to call asymmetric in a stable sense; it changes from sample to sample.

The two cables together describe the same regional dynamic from two different angles. Indonesia and Malaysia have built submarine connectivity between themselves. The routing layer that operates above that connectivity treats both cables as one of many possible paths, and frequently chooses other paths. JaKa2LaDeMa (B2JS), the longer Indonesian-Singaporean ring slightly south of DMCS, shows the same regional pattern: state-built bilateral capacity coexisting with a routing reality that treats Singapore as the gravitational centre of the entire region.

What 39× and 99 ms-of-variance tell us

For most cables in our monitoring set, the floor multiplier and the variance move together — high multiplier with high variance, or low multiplier with low variance. Gondwana-1 sits at 1.081× with a standard deviation of 0.05 ms because there is one path and it is stable. ARCOS-1 sits at 0.613× with low variance because the chord through Florida is also a stable, well-engineered route.

DMCS sits at 39× with a 99 ms standard deviation because there are several routes and BGP is choosing among them in ways that vary over short timescales. The cable is functional. The cable is owned by state operators on both sides. The cable is underutilised — not because of any technical or commercial failure of DMCS itself, but because the surrounding peering fabric makes Singapore the cheaper and more convenient choice for most packets, most of the time.

The Strait of Malacca is, by any maritime standard, the busiest narrow water in the world. The cable across it is, by our measurements, one of the quieter pieces of intercontinental infrastructure between two state operators that explicitly built it together. Both observations are simultaneously true.