Outsourcing is not “borrowing a person”. What you actually buy is two things: a bounded piece of work and a result you can sign off on. Where the boundary is unclear, it ends up as “gear racked but no one dares cut over”, or “cut over, something failed and no one knows who to call”.
First separate the three delivery modes
| Mode | Who is on site | Suitable for | You prepare |
|---|---|---|---|
| Outsourcing project | We send someone on site | Racking, cutover, site move, with clear start and end | Site, time window, change approval, acceptance criteria |
| Remote-hand | Your people on site, we remote | Room does not allow outside access, cross-region, only technical guidance needed | Remote access, a pair of hands on site, a call channel |
| Advisory and review | Not necessarily on site | Existing-network health check, design review, setup review | Existing config, topology, device list |
The same task can be done in different modes. Choosing the wrong mode usually costs not money but time and risk.
Scope: six categories, not all finished in one pass
Installation and baseline setup
Unboxing, racking, cabling, power-on, baseline setup (management access, time sync, logging, backup). It sounds simplest, but this category most needs “a record when done”: port table, serials, config backup, change list — handed over counts as complete.
Cutover and migration
Swap gear, swap segments, move sites. The value here is not “technical difficulty” but window control: pre-validation, the night’s steps, failure rollback, post-check. We do not take cutovers without a rollback plan as routine work.
Routing and VPN
BGP peering (including multi-homing, path adjustment), IPsec VPN tunnels, site interconnect. What to clarify here is “who manages the far end”: carrier, cloud, or another SI. The far end is not a part we control; the liability boundary is drawn up front.
Automation and monitoring
Config templating, bulk push, scheduled backup, config diff comparison; monitoring and alerting onboarding (traffic, device status, alert convergence). The tool layer commonly uses scripts and open-source options; which one actually fits depends on your environment and existing habits — we do not prescribe a stack and ask you to follow.
Branch and wireless
AP onboarding, controller registration, basic coverage confirmation, guest and internal network segmentation. On coverage we are honest: what can be confirmed on site; the depth of coverage planning and RF tuning depends on site and requirements, and is not waved through with a “guaranteed full coverage” line.
Takeover and health check
Take over a network left by others, or prepare to change maintenance vendor. The approach is to read existing setup, organise topology, list risks and to-dos, and hand over a list you can actually use — not a polished slide deck.
Four conditions for remote-hand
Access method
Jump host, management VPN, out-of-band management (console server／KVM) — at least one path that can get in. Access left to “figure out later” at the time equals no access.
Hours and window
When your business can tolerate downtime or slowdown must be stated up front. Remote work also needs a window; it is not “anytime”.
Someone on site
The premise of remote-hand is a pair of hands on site: plug cables, swap modules, press power, watch lights. This person does not need to understand setup, but must follow instructions and report what they see in real time.
Rollback and backup
Back up config and confirm the rollback path before touching anything. This step matters especially in remote-hand — if something goes wrong, we are not on site; you rely on following steps to bring the gear back to the previous state.
Missing one of the four, we suggest switching to an outsourcing project or filling the condition first, rather than forcing it through.
Who has decision rights: how a change proceeds
- Propose: change purpose, impact scope, risk and rollback method.
- Assess: affected systems, time window, who needs to be present.
- Approve: signed off by your lead, not decided unilaterally by us.
- Execute: follow the approved steps; stop immediately anything out of scope.
- Acceptance: confirm against acceptance criteria, archive the record.
The point of these five steps: when something fails, you know which step it came from, rather than blaming each other. We do not “tweak it along the way” without approval.
Scope and liability boundary
| Item | Us | You |
|---|---|---|
| Build steps and technical judgement | Responsible | Provide existing-network data and constraints |
| Change approval and downtime decision | Propose recommendation | Decide and sign off |
| Far-end coordination (carrier／cloud／other SI) | Cooperate | Lead, or authorise us to represent |
| Hardware, licence, third-party service cost | Quoted separately | Decide procurement |
| Site, access, security, power | Cooperate on arrangement | Responsible |
| Acceptance criteria | Recommend | Confirm |
This column is not about shifting blame; it avoids “assuming the other party would do it”. The most common outsourcing disputes come from this column being left unwritten.
Lead time, hours and response
- On-site hours, remote support hours and response method are agreed in the quote and terms.
- We do not preset 24/7 standby, and do not commit to response times outside the contract; night or holiday windows are stated together at quoting stage.
- A project has start and end dates, but those dates depend on your equipment arrival, site and approval process; delays outside our control affect the schedule.
How a quote is built
- By scope: work content, number of sites, number of windows written clearly; suits projects with a defined deliverable.
- By person-day: for floating scope that needs to be defined as work proceeds.
- By phase: health check, design, implementation, acceptance quoted in stages; suits larger or phased projects.
Hardware, licences, third-party services and travel／accommodation are quoted separately, not bundled into a “all-in price” that misleads you. Actual figures follow your list and window arrangement.
When outsourcing fits, and when it does not
Fits:
- Cutover or move with a clear window, with no internal person available at night.
- Branch or room with no resident IT, only occasional on-site visits needed.
- Short-handed during a project period, needing a stretch of execution capacity.
- Taking over someone else’s network, needing one neutral health check.
- Cross-border site with no local person, needing remote-hand support.
Does not fit:
- Only looking for someone to “have a quick look”, with no scope and no acceptance.
- Asking to outsource both design and operational responsibility, but unwilling to provide existing-network data.
- Asking for a fixed price while scope keeps changing.
These three usually exhaust both sides. Talking it through before starting is better for everyone.
Next step
For a quote, email these five items to [email protected]:
- Number of sites and their regions.
- Device models and quantity (or send the list directly).
- Work type (install／cutover／VPN／automation／health check).
- Acceptable window hours, and whether remote work is acceptable.
- Acceptance criteria, or whether you want us to propose a set.
If you only want to ask one question first — “is this worth outsourcing?” — write in directly.
Service scope and liability: this page states the service scope; actual work content, windows, acceptance criteria and liability allocation follow the quote and the terms signed by both parties. Changes execute only after approval by your lead; far-end (carrier, cloud, other contractor) coordination and liability are clarified before the project starts. Export controls, cross-border transport and local regulatory requirements are handled case by case.