China & CERNET 1998: From a Single SprintLink Circuit to Tier 1 Global Internet Peering
Context for readers
This page documents CERNET’s 1998 transition from limited academic Internet access to full global Internet peering. The shift was made possible through Digital Island’s autonomous, multi-continent infrastructure.
It distinguishes China’s early SprintLink connection, which was a single frame relay circuit with domestic transit constraints, from what came next. Full BGP and DNS peering enabled end to end SSL, real time services, and commerce grade performance. As a result, CERNET gained direct reach to 95 percent of the world’s Internet population, virtually overnight.
The financial and technical specifics below are included for precision. Global Internet infrastructure did not appear by accident. It was financed, provisioned, and physically delivered by private capital.
CERNET’s Internet reality prior to 1998
By the mid 1990s, China represented 18 percent of the world’s population, approximately 1.3 billion people.
Yet CERNET’s international Internet access was constrained to a single 64 kbps circuit terminating on SprintLink in Los Angeles. Professor Xing Li of Tsinghua University confirmed this directly to me in Beijing.
Domestically, CERNET linked about 300 universities across China. Internationally, it was dependent on that lone upstream line. One phone circuit. No peering. No symmetry.
CERNET could access the Internet. But it could not peer, could not route, could not host, and could not participate in the global exchange of traffic on equal terms. It was present, but not sovereign.
The SprintLink constraint
SprintLink offered basic transit. CERNET’s international traffic exited into Sprint’s backbone through a standard customer provider arrangement. Routing control, performance optimization, and policy enforcement were outside CERNET’s reach.
The line was frame relay. It was not designed for SSL, predictable latency, or large scale bidirectional traffic. The architecture was suitable for email and text retrieval, not for commercial, secure, or real time use.
This was not global Internet presence. It was narrowband dependence on a single U.S. carrier.
Digital Island’s intervention
In February 1998, I traveled to Beijing and worked directly with Professor Xing Li and his team at Tsinghua University to build a solution.
The goal was to shift CERNET from a dependent access node to a global Internet peer with full route control, DNS symmetry, and secure interconnection to the world’s major ISPs and content hubs.
We designed and deployed a direct interconnect and autonomous route exchange between CERNET and Digital Island’s Tier 0 backbone. This removed the SprintLink choke point.
No one else was willing to provision this infrastructure. So we did.
The financial reality
Digital Island provisioned submarine and terrestrial capacity across the Pacific. The link was dedicated exclusively to CERNET. Digital Island provisioned an International Private Line Circuit (IPLC) to support autonomous global peering.
Costs:
-
Annual circuit cost: $960,000
-
Prepaid upfront: $275,000
-
One time install fee: $35,000
-
Monthly recurring charges: $80,000
-
Three months prepaid: $240,000
CERNET paid nothing. No Chinese entity funded this transition. The entire cost was borne by Digital Island, directly from my cost center.
This was risk capital, not billable bandwidth.
The technical shift
The new circuit delivered a T1 at 1.544 Mbps, which was 24 times the speed of CERNET’s 64 kbps SprintLink line.
But the breakthrough was not speed. It was control.
This was dedicated point to point international capacity. It was not shared, not oversubscribed, and not dependent on a U.S. carrier’s domestic routing logic.
Digital Island installed routing and content infrastructure directly inside CERNET’s Beijing footprint. Full BGP peering. DNS authority. Autonomous global route exchange.
Latency moved into a different class. CERNET gained sub 300 millisecond round trip reach to the majority of the commercial Internet, with routing autonomy and end to end SSL viability. This was not theoretical. It was operational and measurable.
That combination of IPLC transport, global route exchange, and low latency is what turned CERNET from a constrained academic access network into a commerce grade participant in worldwide Internet traffic.
What changed immediately
After the routing cutover and DNS updates, CERNET gained direct access to:
-
Six continents
-
Major global ISPs
-
Nineteen Digital Island data centers
-
A commercial scale content distribution platform
Users experienced immediate, obvious change. Not just faster downloads, but predictable global behavior.
CERNET gained:
-
Dedicated IPLC capacity into a global backbone
-
Full bidirectional BGP route exchange and DNS symmetry
-
Sub 300 millisecond round trip paths across major global markets
-
End to end SSL capable sessions suitable for secure commerce
-
Direct reach into Digital Island’s worldwide infrastructure footprint
This is the difference between “reachable” and “usable at global scale.”
Why this mattered
This was not an academic upgrade. It was an economic unlock.
Digital Island already carried commerce and enterprise platforms that required predictable global performance and secure sessions. That included Visa, MasterCard, E*TRADE, Charles Schwab, Microsoft, Stanford, Cisco, and Intel.
Once CERNET was peered into our IPLC backed, sub 300 millisecond global network, those platforms had a commerce grade path into mainland China. China could participate on equal network terms, and global services could enter China with real performance and real security.
That is what “globalization of eCommerce” looks like at the infrastructure layer.
Attribution clarification
SprintLink provided basic upstream access to CERNET. It did not provide global peering, routing autonomy, or predictable worldwide performance. It provided USA-centric, constrained reachability through a single carrier path.
Digital Island provided Internet parity: bidirectional BGP route exchange, DNS symmetry, sub 300 millisecond round trip performance across major global markets, and SSL viable end to end sessions at global scale.
CERNET did not fund this transition.
Digital Island did not invent the Internet.
Tier 0 is used here to describe an autonomous, multi-continent backbone built as an overlay using IPLCs and other dedicated private circuits across multiple underlying Tier 1 carriers. It was designed for direct peering, global route control, and deterministic performance across regions, independent of any single Tier 1 provider’s transit constraints. This was Digital Island’s core value proposition, and it is what CERNET gained in 1998.
What Digital Island delivered was the first autonomous, Tier 1 equivalent Internet presence into mainland China, with full BGP route control, DNS symmetry, SSL capability, and direct interconnect to every major global region.
This was Internet parity. No one before Digital Island had financed and deployed an autonomous, commerce-grade peering link that gave CERNET global routing parity, sub 300 millisecond round trip performance across major markets, and SSL viable end to end connectivity at scale.
I make this statement as a matter of direct personal knowledge. I was the only person on the ground executing this peering deployment with Professor Xing Li and his team. The full cost was borne by my cost center at Digital Island. Digital Island performed the provisioning, installation, and routing cutover. There was no other sponsor or parallel implementation delivering this result at that time.
Any contrary claim should identify the alternate funder, the alternate circuit, and the dated cutover record.
Correction note
An earlier email misstated the timing of my Beijing visit. The correct date was February 1998.
A final note on capital
This link was not funded by government, academia, or Chinese institutions. It was paid for by private capital.
Cliff Higgerson did not invent the Internet. He financed the infrastructure that made it globally usable.
See: https://marknichols.com/cliffhiggerson/
Claim for recognition
This was not a prototype. It was operational. It was global. It was live.
The path between China and the Internet became bilateral, secure, and equal because of this deployment. If global Internet parity matters, this should be recognized as one of the defining infrastructure moments in Internet history.
This page is filed as Evidence Node 5 in the Digital Island Evidence Vault.