Cloud connect and multicloud
direct to AWS, Azure, GCP and Alibaba Cloud
Connecting an office or a rack to public cloud has three possible answers, and they differ by an order of magnitude in cost. The wrong one usually reveals itself only after traffic grows.
What this service covers
Direct public-cloud connect
Physical connectivity into AWS, Azure, GCP, Alibaba Cloud and other major providers, bypassing the public internet.
Multicloud on one fabric
Cloud regions join the same network and route through it directly, instead of hair-pinning back through your premises.
YUNSHU self-service
VPC, routing, NAT, egress and inter-cloud links selected self-service and live within minutes.
Hybrid designs
Racks, offices and cloud regions planned as a single topology with consistent addressing.
Redundancy options
Dedicated capacity for daily traffic with VPN standby, switching automatically on failure.
Operations
24/7 NOC monitoring, proactive alerting and a visual portal across the whole fabric.
Who uses it, and for what
Multicloud interconnect
- Direct connect to AWS / Azure / GCP / Alibaba Cloud
- Route convergence handled network-side, not via your premises
- YUNSHU self-service, changes live in minutes
- Delivered in 3 working days
Peak elasticity
- 10 Mbps – 100 Gbps, upgradable in steps
- Campaign peaks billable from one day
- Colocation with three-carrier bandwidth
- 99.5% end-to-end availability
Plants, colocation and cloud on one fabric
- Deterministic transport for production systems
- Cloud regions joined to the same network
- Analytics workloads land where they run best
- 24/7 NOC monitoring
Overseas cloud regions
- International segments over the three carriers' licensed networks
- Application-aware steering around peak congestion
- Interconnect with overseas IDC and cloud regions
- Flexible terms, from one day
Service parameters at a glance
These are network-wide figures. Anything that depends on a specific site — facility conditions, access method, exact schedule — is confirmed by survey.
| Item | Specification |
|---|---|
| Clouds reachable | AWS · Azure · GCP · Alibaba Cloud and other major providers |
| Lead time | Cloud connect delivered in 3 working days; additional time if a new access segment is needed on your side, confirmed after survey |
| Self-service | With the YUNSHU platform, VPC, routing, NAT, egress and inter-cloud links are selected self-service and live within minutes |
| Recommended topology | Join cloud regions to one fabric and converge routes there, so cloud-to-cloud traffic never hair-pins through your premises |
| Redundancy | Dedicated capacity for daily traffic with VPN standby, switching automatically on failure |
| Billing | Fixed monthly rate (monthly / quarterly / annual); installation listed as a one-off item. Cloud-side port, virtual-interface and egress charges are billed by the cloud provider under their own rules |
| Bandwidth | 10 Mbps – 100 Gbps, upgradable in steps |
| Service assurance | PoP-to-upstream-ISP availability 99.99%; end-to-end link availability 99.5%; 24/7 NOC monitoring and proactive alerting; 100% VIP fault tracking |
| Contract term | Flexible — no blanket minimum. Annual terms for steady services; short-term needs (events, exhibitions, temporary projects) from one day |
| Footprint | 60+ PoPs across Greater China, 120+ PoPs worldwide; measured latency between nodes is queryable on the coverage map |
How to think about it
Public internet, IPsec VPN, or cloud connect. The internet costs nothing extra and guarantees nothing. A VPN adds encryption and unified addressing on top of the internet — it fixes security and IP planning, not quality; the tunnel still rides the same congested path, and adds crypto overhead. Cloud connect gives dedicated bandwidth and predictable latency, at the highest cost and with a real lead time.
Two business questions usually settle it. If peak-hour bandwidth halved, what happens? If the answer is "users complain" or "jobs fail", you need a guarantee. How much data actually crosses between premises and cloud each day? Past a certain volume, egress charges plus the experience penalty exceed the fixed cost of a dedicated connection — and that crossover arrives sooner than most teams expect.
Do not let cloud-to-cloud traffic hair-pin. If each cloud gets its own line back to your premises, traffic from cloud A to cloud B crosses your site twice: latency doubles, your egress becomes the bottleneck, and a local failure cuts cloud-to-cloud connectivity. Join the cloud regions to one fabric and let route convergence happen there. Fredyun delivers cloud connect in 3 working days; with the YUNSHU platform, VPC, routing, NAT, egress and inter-cloud changes are self-service and live within minutes.
Licence scope
International and Hong Kong / Macau / Taiwan capacity is not built by Fredyun. China Mobile International (CMI), China Unicom Global (CUG) and China Telecom Global (CTG) hold the international telecom licences and supply the underlying circuits. Fredyun procures that licensed capacity and delivers convergence, routing, end-to-end provisioning and operations on top of it.
Fredyun itself holds a trans-regional value-added telecom business licence (B1-20235481) issued by China's Ministry of Industry and Information Technology, covering IDC, ISP, CDN and domestic internet VPN — trans-regional, not international. This boundary is stated explicitly in every proposal and contract.
What clients ask most
Is egress still charged over a dedicated connection?
Usually yes. Per-gigabyte egress over a dedicated connection is normally cheaper than over the internet, but it is still billed. When planning, size both steady-state egress and the exceptional cases — full backups, DR drills, data migrations, bulk log shipping — because those peaks are often tens of times steady state. Set billing alerts and schedule bulk transfers. Exact rules are set by each cloud provider.
What is the right way to connect several clouds?
Join the cloud regions to one fabric and let the network converge routes, so clouds reach each other directly. Avoid giving every cloud its own line back to your premises — that hair-pins traffic, doubles latency and makes your egress the bottleneck. Cross-cloud topology and address planning are settled during solution design.
How quickly can cloud connect be delivered?
Three working days on the Fredyun side. Cloud-provider approval and configuration time is separate and varies by provider. If a new access segment is needed at your premises, that is added and confirmed after the link survey — local works and the cloud-side request can run in parallel.
Can dedicated connectivity and VPN be combined?
Yes, and it is a common redundancy pattern: the dedicated connection carries daily traffic while a VPN stands by and takes over automatically if it fails. You get the determinism without buying a second dedicated circuit. Failover policy and detection are set during solution design.
Usually planned together
Need a link assessment for this design?
Tell us the site locations, bandwidth and latency targets
and we will come back with a workable topology and SLA proposal.
