Batam-Rengit Cable System (BRCS)

Based on 1,023 measurements from GeoCables monitoring infrastructure, March-April 2026.

The Batam-Rengit Cable System - BRCS - is a sixty-four-kilometre submarine link between two points that face each other across a narrow stretch of the Singapore Strait. The Indonesian landing sits at Tanjung Pinggir, on the northern coast of Batam Island, the largest island in the Riau Archipelago. The Malaysian landing is at Rengit, a small coastal village in the southern Johor district of Batu Pahat, fifteen kilometres south of Singapore's land border with Malaysia. The cable was placed in service in 2007, has been operational for nineteen years, and is owned by XLSmart - an Indonesian operator formed in 2025 by the merger of XL Axiata and Smartfren. By any meaningful submarine standard, sixty-four kilometres is a short cable. The Atlantic crossings are 6,000 km. The Pacific crossings exceed 13,000. BRCS is the cable equivalent of a footbridge between two municipalities of the same metropolitan region.

A footbridge that, judging by our measurements, almost no traffic uses.

Light in optical fibre travels at roughly two hundred thousand kilometres per second - about two-thirds of vacuum c. A round-trip across sixty-four kilometres of glass should resolve in 0.63 ms. That number is the physics floor for BRCS: the smallest possible round-trip, the value below which no measurement on this cable can fall without violating relativity. We do not measure 0.63 ms. We measure 50.40 ms. The minimum round-trip we have observed in 169 measurements between Rengit and Tanjung Pinggir, sent in the standard direction MY → ID, is exactly that: fifty milliseconds and forty microseconds. That is eighty times the physics floor. For a cable this short, an 80× multiplier is not measurement noise. It is a routing decision rendered as a measurement.

Forward and reverse take different oceans

The story sharpens when both directions are inspected. Across 169 measurements from Rengit (MY) to Tanjung Pinggir (ID), the average round-trip is 278.75 ms. Across 20 measurements in the opposite direction - Tanjung Pinggir to Rengit - the average is 112.97 ms. Both directions involve the same physical pair of endpoints. Both should produce the same measurement on a 0.63 ms-floor cable. The forward path takes 2.5× as long as the reverse, and both are far above the floor. The symmetry gap between them is roughly 198 ms.

This kind of asymmetry is not a property of the cable. Submarine cables themselves do not introduce direction-dependent latency at this scale; light travels at the same speed in both directions of a fibre pair, and the wet plant - repeaters, branching units, terminal equipment - is symmetric by design. The asymmetry is introduced in BGP. Each side of the connection has independent control over how its outbound traffic is routed, and that routing is governed by transit costs, peering agreements, and AS-path length. The existence of a particular submarine cable underneath the route is not a parameter the BGP decision algorithm sees.

What our measurements describe is not BRCS's link characteristics. They describe the routing system around BRCS, observed through the cable's terminal IPs. The MY-side operator's outbound preference takes traffic out to a regional carrier hub - most plausibly Singapore, which sits within twenty kilometres of both landing points and hosts every major Asia-Pacific carrier exchange - before that traffic returns south to Batam Island. The ID-side operator's outbound preference takes a shorter detour, but still not the cable directly. The cable is in the water, electrically powered, optically transmitting. It is also, by the routing layer's reckoning, irrelevant.

Why a sixty-four-kilometre cable exists in the first place

If neither side is using BRCS for direct ID-MY exchange, why does the cable exist? The answer is the same answer as for most short submarine cables in the region: it is cheaper than the alternative, for some specific kinds of traffic, even if it is not the chosen path for general internet exchange. BRCS provides a bilateral backstop. It allows XLSmart's Indonesian customers to reach Malaysian destinations without paying transit through Singapore IXPs for a subset of traffic - typically operator-internal flows, capacity reservations, or specific content-delivery agreements where the two carriers have negotiated bilateral terms.

The general internet, however, follows the BGP path. And that path goes through Singapore.

Where the variance lives

Long-distance probes targeting Tanjung Pinggir from Moscow, Minsk, Tbilisi, Jerusalem, Sevastopol, and Almaty produce strikingly consistent round-trips. The 138 measurements from Minsk show a standard deviation of 0.81 ms across an average of 207 ms. The 140 measurements from Moscow show a standard deviation of 0.84 ms across an average of 207.68 ms. These are international paths, ten thousand kilometres long, transiting four or five separate cable systems and as many peering points, and yet they are stable to within a millisecond. The 169 measurements over the local pair - sixty-four kilometres - have a standard deviation of 68.16 ms.

The cable that should be the cleanest measurement on the system is the noisiest one. International long-haul routes have been ironed flat by twenty years of peering negotiations between major Tier-1 carriers; domestic and bilateral edges between two adjacent national operators carry the residual jitter of inter-carrier policy, traffic engineering experiments, and the soft economics of capacity provisioning. We have measured the same shape on JaKa2LaDeMa (B2JS), the longer Indonesian domestic ring that terminates a few hundred kilometres from BRCS. Its local pair shows the same elevated variance and the same routing-driven floor multiplier.

What 80× looks like next to other cables

For most cables in our monitoring set, the observed minimum round-trip lands somewhere between 1× and 5× the physics floor. Gondwana-1 between New Caledonia and Sydney measures at 1.081× - essentially at the floor - because it is the only cable carrying that traffic, with no realistic alternative path; routing has nowhere else to go. ARCOS-1 in the Caribbean measures at 0.613× - below the floor - because traffic chords across the Caribbean basin via Florida hubs instead of looping around the cable's full ring path.

BRCS at 80× sits at the opposite end of this spectrum. The cable exists. The cable is operational. The cable is owned by an operator with a substantial subscriber base on at least one of its sides. The routing layer surrounding the cable does not use it. Three out of every four packets travelling from Rengit to Tanjung Pinggir spend their time in carrier networks that are neither Indonesian nor Malaysian - Singapore being the most likely waypoint. The cable is sixty-four kilometres. The route is not.

The 80× floor multiplier on BRCS is, in the end, a quantitative summary of the gap between owning a cable and being routed across it. The first is a capital expense decision. The second is a routing-policy decision. The two do not always agree, and when they do not, the cable shows up in our measurements as a fast direct link that almost nobody travels.