The client and the starting point
What makes this group unusual is that its sites are not the same thing at all. A logistics warehouse has to survive forklifts and steel racking; a cosmetics store has to carry live streaming and a loyalty system; the pharmaceutical arm has to pass compliance audits; trading offices have to host video calls across several regions. Four business units, four sets of requirements, all held to one standard.
Until then the network was maintained in pieces: a local integrator looked after one part, the vendor another, head-office IT handled change. The result was nobody owning it end to end — when something broke, the first task was agreeing who was responsible, not finding the fault. What the client wanted was simple: one counterparty that could be held to a score.
Seven problems on the ground
The difficulty in work like this is never the technology choice. It is complexity, time zones and measurement all arriving at once.
Cross-region complexity
A hundred offices and thirty thousand devices across multiple vendors, protocols and regions. Any investigation starts by working out whose territory the device is in.
Someone has to be awake at night
00:00–09:00 local time is the working day in Europe and the Americas. The night shift has to cover network management, SD-WAN and project coordination — a forwarded on-call phone is not enough.
SLAs with penalties
Service, asset, incident, problem and change management each carry an explicit SLA and penalty, scored monthly. Missing the target triggers the penalty. This is not a best-effort arrangement.
Four business units, four temperaments
One standard, but room for difference. Force a single template and the business units push back; let them do as they like and cost runs away.
Multi-vendor coordination
Equinix, Fortinet, the carriers and the hardware vendors all have to be engaged at once. During fault isolation, what actually slows things down is waiting for someone to answer an email.
Change must be compliant
Every change goes through the client's CAB: SR raised, TD prepared, reviewed, implemented, quality-checked. No step is skipped.
Documentation cannot go stale
Cross-border circuit inventories, high- and low-level designs and topology diagrams have to match the live network. Most operational incidents trace back to documentation that fell behind reality.
Operating model
Ten mechanisms, each locking into the next, make up a delivery model of 24/7 cover, tiered response, SLA control and a closed compliance loop. They fall into five groups.
Cover and response
Ten people over three shifts give continuous follow-the-sun cover, with the night shift focused on SD-WAN and network management. Tiered L2/L3 support defines the escalation path, so every incident has a clear next owner.
Scoring and review
A monthly service report and SLA scorecard, with an improvement plan for every missed line. Weekly and monthly reviews cover incidents, change, capacity and SLA performance, turning quality into a curve you can track rather than a feeling.
Change and compliance
The client's CAB process is followed strictly, with SR, TD, review, implementation and quality check all controlled and traceable. Emergency changes take a fast lane and are documented afterwards.
Assets and documentation
A cross-region asset inventory, topology diagrams and high- and low-level designs, refreshed on a schedule and shared with the client. SSL certificates, circuit inventories and physical asset checks are part of the annual routine.
Coordination and knowledge
A named vendor liaison maintains fast-response channels with Equinix, Fortinet and the carriers. Knowledge transfer completes during transition, and the runbooks and work instructions keep being revised.
Proactive monitoring
Using the client's existing tools plus Penghai's own scripts, devices, links, traffic and authentication are monitored from several angles — finding problems before users report them.
Scope: the whole network lifecycle
Not break-fix, but a closed loop from monitoring and incidents through change and assets to reporting.
After three years
There is no dramatic firefighting story, and that is the ideal outcome. Three consecutive years without a major outage, monthly SLA scores consistently met, and the contract renewed each time.
The more concrete value shows up day to day: when the client opens a new office, the network is no longer a project that needs re-approving but a line that already runs — standard configuration, an established change process, a known delivery lead time. The cost curve of expansion flattens, and that is what long-term operations actually buys.