East-West Submarine Cable System

The East-West Submarine Cable System (EWSCS) is a 950-kilometre regional fibre-optic link operated by Sacofa Sdn Bhd, a Malaysian telecommunications company headquartered in Kuching, Sarawak. Put into service in 2004, the cable connects Mersing on the east coast of Peninsular Malaysia (Pahang state) to Penarik on the eastern coast of Sumatra, Indonesia, with intermediate landing points at Kuching and Terempa.

By any single metric — length, age, ownership, or international relevance — EWSCS is unremarkable. It is shorter than the distance from London to Lisbon, older than most other cables we monitor, owned by a single regional operator, and lands in two minor sites rather than the major hubs at Singapore (Tuas, Changi) or Jakarta. Yet measurement data from RIPE Atlas tells a story that is anything but unremarkable.

The Numbers That Don't Add Up

For a fibre cable of length L, the theoretical minimum round-trip time (the "physics floor") is roughly 2L / (0.667 × c), where 0.667 × c accounts for the speed of light in glass. For 950 km, that floor is approximately 9.3 ms. Any measurement faster than this would imply the signal is taking a shortcut — which generally means we have the cable length wrong, or routing is not actually using the cable.

Across 208 RIPE Atlas samples spanning seven probe origins over the last 30 days, here is what EWSCS looks like:

The pattern is strange enough to merit a second reading. When we measure from the cable's own landing point in Mersing, just 950 km from Penarik, latency varies wildly between 16 ms and 561 ms — a 35× spread. When we measure from Jerusalem, almost 8,000 km away through five hops of trans-Asia infrastructure, the round-trip is rock-stable at 227 ms with a standard deviation of 0.21 ms.

Why Distant Probes Are More Predictable Than Local Ones

This is a familiar paradox in network measurement, but EWSCS makes it especially vivid. The cable itself is 950 km of glass with a fixed forward delay of about 4.7 ms. Once you pay that physical cost, the rest of any RTT comes from access networks, peering decisions, and queue depth at each hop along the path.

From Jerusalem, the path to Penarik is dominated by long-haul trans-Asia trunks (Europe→Egypt→India→Singapore→Indonesia, give or take), each constrained by its own physics floor. The path is long but mostly fibre between large IXes, where queuing is minimal. The result: variance averages out via something like the central limit theorem at the network layer.

From Mersing, however, the path to Penarik is dominated by Malaysian and Indonesian last-mile infrastructure — networks designed primarily for residential and enterprise traffic with deep buffers and significant peak-hour congestion. A 950 km hop touches more "messy" infrastructure as a percentage of total distance than an 8,000 km hop does. Short paths see the messiness up close; long paths smooth it out.

Asymmetric Routing — Same Fibre, Different Return Paths

The 35× variance on Mersing→Penarik strongly hints at asymmetric routing. The minimum measured RTT of 15.76 ms is just 1.7× the physics floor — close enough to indicate that ICMP packets in the forward direction probably do traverse EWSCS as expected. But the average climbing to 227 ms, and a maximum of 560 ms, tells us the return packets from Penarik are taking a very different path.

One plausible explanation: forward packets from Mersing follow a short BGP-preferred route across EWSCS, while return packets from Penarik are sometimes routed through Jakarta or Singapore IX peering, then back across alternate cables, before reaching Mersing. When those alternate paths are uncongested, the round-trip ratio still feels reasonable; when congestion hits, RTT balloons to 500+ ms. The signature is exactly what we see: very low minimum (one direction uses the cable) and very high maximum (the return takes a long detour).

What This Tells Us About Regional Cables

EWSCS is by design a regional asset — Sacofa-owned, terminating at minor coastal sites rather than mega-IXes — and the measurement story it tells is also regional. Most of our other cables in this dossier carry trans-oceanic traffic, where path symmetry is enforced by the small number of viable carriers. Regional cables like EWSCS sit in a denser mesh where alternate paths are abundant, BGP decisions cheap to make, and asymmetry the norm rather than the exception.

The minimum RTT of 15.76 ms tells us the fibre itself is healthy and forward routing is using it. The average of 227 ms tells us the return path is going through something else, somewhere else. Continuous monitoring will reveal whether this asymmetry is stable (BGP picking the same alternate path consistently) or fluctuating (multiple paths chosen across the day). Either reading would be useful.

EWSCS in the Regional Cable Mesh

EWSCS is one of several Sacofa-operated cables linking Peninsular Malaysia, Sarawak, and Indonesian Sumatra. It coexists in the regional mesh with SEAX-1, B2JS, MCS, and shorter domestic cables, all of which provide redundant paths between Singapore-area landings and the broader Indonesian archipelago. The 2004 RFS date places EWSCS in the post-dot-com cable era — cables built for survival rather than the speculative oversupply of the late 1990s — and twenty years of operation have given it ample time to integrate into a complex web of regional carriers.

What makes EWSCS interesting for measurement work is precisely this: a healthy 950 km cable with a clear physics-floor signal in the forward direction, embedded in enough alternate-route infrastructure to produce dramatic asymmetry in the round-trip view. It is a small cable that teaches a large lesson about how submarine fibre actually behaves once it leaves the lab.

Why This Cable Matters for Network Researchers

For practitioners and researchers studying submarine cable behaviour, EWSCS sits in a useful spot. Long trans-oceanic systems make for clean physics-floor demonstrations because their length swamps any access-network noise. Very short fibres are dominated entirely by access. EWSCS at 950 km lies near the boundary where both effects matter equally — making it a sensitive probe for how regional carriers in Southeast Asia actually wire their inter-network handoffs. The contrast we see between local and remote measurement origins is, in that sense, a tutorial in why distance and predictability often have an inverse relationship in real networks.