Network equipment · switching and routing
Full-rack supply and integrated delivery
From core to access, from selection to acceptance — what is installed in each rack, the bandwidth served by each rack unit, and where each item appears in the delivery list, all specified at quote stage. No approximate supply: count every device against the drawings at acceptance.
This is a rack we deliver
A standard 42U rack, from U42 to U1, with every device identified by number, model and function. Every line in the quote maps to a rack unit in this drawing — On arrival, check each device against the drawing: serial number, firmware version and port count. Sign only after every item is verified.
Cabling and topology sketch
CABLE ROUTINGSix major product categories, from hardware through compliance
Itemized by category, byBrandandModel rangeDefine capability boundaries. Verify channel status, pricing and support scope for each item during procurement. The product lines below are selection references; confirm each item during commercial review.Supply-channel source, warranty start date and manufacturer support ID, verify directly with the manufacturer after equipment arrives.
Campus and data-center switching
SWITCHINGRouting and WAN connectivity
ROUTING / WANWireless coverage
WLAN / WIRELESSNetwork security and zero trust
SECURITY / ZTNABastion host and operations audit
BASTION / AUDIT · 4A compliance and MLPSOptics and cables
OPTICS / CABLINGCredentials and supply channels:Manufacturer-issued documents showing authorization numbers, validity and coverage can be provided for review during procurement. No gray-market supply or verbal promises; the quote specifiesSupply-channel source, warranty start date and manufacturer support IDThree checks; after delivery, verify directly with the manufacturer.
Four-layer architecture: one concern per layer
Campus-network complexity is not about the number of devices, butResponsibilities are mixed together— Running policies on the core and routing at the access layer makes faults hard to isolate. Our default design separates the layers: each layer forwards traffic while policy and redundancy stay in their proper places.
Perimeter layer
- Equipment form factor1U–2U rack-mountable, dual power supplies
- Port density8×10G SFP+ + 8×GE copper/fiber combo ports
- UplinkDual-carrier private circuits,1G / 10G
- RedundancyHA pair, primary/standby failover < 3 seconds
This layer solves one problem:A single WAN circuit is a single point of failure. Two private circuits plus an HA pair are the minimum configuration at this layer.
Core layer
- Equipment form factor1U / 2U fixed or chassis form factor
- Port density32×100G QSFP28, supports 1-to-4 breakout
- Uplink25G / 100G uplinks to aggregation
- RedundancyDual-unit stack with separate control and forwarding planes
This layer solves one problem:Forwarding must be fast and continuous. Do not attach endpoints or apply policy at the core: a core policy fault can affect the entire building.
Aggregation layer
- Equipment form factor1U rack-mountable, stackable
- Port density48×10G ＋ 8×25G Uplink
- UplinkDual 25G / 40G uplinks to dual core switches
- RedundancyDual-homed across core switches with link aggregation
This layer is the policy boundary:Inter-VLAN routing, ACLs and QoS are implemented here. Centralizing policy makes the impact of a change clear.
Access layer
- Equipment form factor1U rack-mountable, optional low-noise model
- Port density48×1G PoE++ ／ 24×2.5G PoE++
- UplinkDual 10G, dual uplinks across aggregation
- PowerPoE / PoE+ / PoE++; budget total system power at 1.3×Core
This layer sees the most frequent failures:Ports, cabling and power all come together here. Therefore, plan locations by 15%–20% Reserve spare ports and 30% PoE power above full-load demand.
Three configurations: choose redundancy first, then brand
The same device locations can produce quotes that differ threefold. The difference is not the unit price, butRedundancy and service tiers. We recommend selecting the tier first and the brand second: the tier determines the architecture, while brands are alternatives within that tier.
Standard
One office building or factory floor, up to 200 data points, where several hours of disruption are acceptable.
- Single core switch,1G At your desk
- Aggregation by floor, with dual 10G uplinks
- Access · 48×1G PoE+, with 15% spare capacity based on device locations
- Single uplink; recovery requires manual failover
- Manufacturer NBD Spares, 8×5 ticket support
- Complete acceptance documents, including topology and port maps
High-availability tier
Multiple campus buildings or high production-continuity requirements, 200–800 data points, with outages measured in minutes.
- Dual-core stack, separate control and forwarding planes
- Dual aggregation switches, dual-homed across core, with link aggregation
- Backbone 25G, 10G from aggregation to access
- Access · PoE++; budget power at 1.3× full load
- Manufacturer 8×5 TAC,Next-day spare-parts dispatch
- Includes configuration backups, emergency plans and semiannual inspections
Active-active disaster recovery
Primary/backup server rooms or two-site/three-center designs, where outages are costly and failover must be testable.
- Two server roomsIndependent core for each network, 40G / 100G interconnect
- Layer 2 and gateway redundancy across server rooms, with testable failover
- Backbone 100G, dual uplinks end to end, with no single point of failure
- HA pair at the security perimeter, with intrusion and encrypted-traffic inspection
- Manufacturer 24×7 TAC，4 hoursEngineer on site
- Annual disaster-recovery exercise and configuration-baseline review
Illustrative estimate range:200 data points, domestic brands, implementation and first-year maintenance included. International brands, 100G optics, long-reach single-mode fiber and on-site service can materially increase cost. Final pricing is calculated item by item in the BOM, not as a vague bundle.
How many data points are there?
This determines the number of access switches and spare ports. Reserve 1.15–1.2× the actual port count to avoid adding switches next year.
What bandwidth does the backbone need?
This determines core and aggregation uplink optics and fiber type (OM4 multimode / OS2 single-mode), and whether recabling may be needed within five years.
Is PoE power required?
This determines access-switch power, rack circuits and cooling. Budget 1.3× the full endpoint load to prevent intermittent camera and AP outages.
How much downtime is acceptable?
This determines redundancy, spares and manufacturer service levels. Accepting a four-hour recovery versus requiring recovery within one minute means very different designs and quotes.
Eight phases, each with a signable deliverable
The most disputed part of hardware delivery is “what counts as complete.” For that reason, each phase is defined bya deliverable that can be signed off— You are not signing off on a “status update,” but on a document that can be checked and archived.
Site survey and device-location confirmation Both parties
Confirm the low-voltage room, available rack units, power circuits and cooling on site, and compare drawings with reality. Without this step, every quote is a guess.
Equipment selection and BOM quote Confirm per project
Provide three options and an itemized BOM listing model, quantity, unit price, warranty and lead time. Compare other quotes against the same BOM; we do not rely on information asymmetry.
Manufacturer ordering and stock allocation Our team
Order through the agreed supply channel. Verify purchase records, serial numbers and support coverage against the actual documentation. Confirm lead times in writing by brand, model and stock availability.
Inspect equipment on arrival Both parties
Unbox together and check the serial number, model, firmware version and accessories for every device. Photograph and archive the evidence before signing; this prevents many later disputes.
Rack installation and structured cabling Our team
Plan equipment by rack unit, manage cabling, label devices and terminate patch panels; test each fiber after splicing. Labels match drawing references so the next team can understand the rack years later.
Configuration and integration testing Our team
Configure VLANs, Layer 3 gateways, routing protocols, stacking, link aggregation, ACLs and QoS policies item by item, after backing up the factory configuration.
Load and acceptance testing Joint acceptance
Test each acceptance item: port throughput, redundant-link failover time, routing convergence, long-duration packet loss and full-load PoE power. Fix failures before sign-off; do not accept based on “it feels fast.”
Documentation handover and operations training Both parties
Hand over complete documentation and provide on-site training: reading topology, checking ports, backing up configurations and knowing whom to contact. The goal is for your team to take over confidently.
What you are buying: five categories, 21 verifiable line items
The most common pitfall when buying network equipment isEquipment-only quote, excluding licenses and implementation— Discovering after delivery that advanced features require separate licenses or cabling requires another contractor creates surprises. This checklist forms the quote, with every line carried into the contract appendix.
| Category | Project and specifications | Quantity basis | Warranty / term |
|---|---|---|---|
| A · Hardware | |||
| A-01 | Core switchDual units with redundant power and fans; choose 32×100G or 24×25G based on tier | By architecture tier | 1 year for the appliance, extendable to 5 years |
| A-02 | Aggregation switch48×10G + 8×25G uplinks, dual-homed across the core | By number of aggregation nodes | 1 year for the appliance, extendable to 5 years |
| A-03 | Access switch48×1G or 24×2.5G, PoE+ / PoE++; budget power at 1.3× full load | Number of data points ÷ 48 | 1 year for the appliance, extendable to 5 years |
| A-04 | Wireless APs and controllerWi-Fi 6 / 6E, powered by PoE or local power | By coverage area and concurrent users | 1 year for the appliance, extendable to 5 years |
| A-05 | Router / next-generation firewallHA pair with redundant power and high-availability licenses | By WAN egress and number of zones | 1 year for the appliance + security subscription |
| A-06 | Optics and interconnect cablesSFP+ / SFP28 / QSFP28, DAC / AOC; test every optical module | 1:1 port capacity + 5% spare parts | 1 year, test report included |
| A-07 | Rack and accessories42U rack, PDU, cable managers, copper and fiber patch panels | By rack count | Rack and accessories · 3–5 years |
| B · Licenses and subscriptions | |||
| B-01 | Switch software licensesAdvanced Layer 3 features, stacking / virtualization, SDN and automation interfaces | By device count | 1 / 3 / 5-year term options |
| B-02 | Wireless licensesController and AP licenses, including roaming and RF optimization | By number of APs | 1 / 3 / 5-year term options |
| B-03 | Security subscriptionIntrusion prevention, antivirus, URL filtering and threat-intelligence updates | By firewall count | 1 / 3 years, renewed annually |
| B-04 | Network and cloud-management platformCentralized management, configuration backups, alerts and reports | By managed-device count | Annual subscription |
| C · Vendor services | |||
| C-01 | Manufacturer support services8×5 or 24×7 TAC, software and firmware upgrades, and RMA channel | By equipment list | 1 / 3 / 5 years, renewable |
| C-02 | Spare parts firstNext-business-day delivery or four-hour engineer arrival, by service tier | By service tier | For the maintenance term |
| C-03 | Locally stocked spare partsCommon optics, power supplies and fans; full-device spares for critical nodes | 5% or as agreed in the contract | Create inventory records at handover |
| D · Implementation | |||
| D-01 | Structured cablingCat6 / Cat6A copper, OM4 multimode or OS2 single-mode fiber, including termination and link tests | By data-point count and cable-run length | Manufacturer-backed cabling warranty: 15–25 years |
| D-02 | Rack installation and cable managementPlan rack units, mount and secure equipment, manage cables and label devices to match drawing references | By rack count | Workmanship warranty: 1 year |
| D-03 | Fiber splicing and testingSplicing, termination, optical-loss and OTDR testing, with a report for every fiber | By fiber count | Test report included with delivery |
| D-04 | Legacy-network cutoverCut over during a maintenance window, with a rollback plan and full pre-cutover configuration backup | Per occurrence | Includes rollback safeguards |
| E · Documentation and training | |||
| E-01 | As-built documentationTopology · rack elevation · port map · configuration backup · acceptance report | 1 set, including editable source files | Valid long term |
| E-02 | Operations manualInspection checklist, incident severity and escalation path, emergency-response procedure | 1 set | Valid long term |
| E-03 | On-site trainingFor IT and operations teams, with hands-on port checks, configuration backups and incident classification | 1–2 on-site sessions | During the delivery period |
Incident severity and on-site response times, written into the contract
“7×24 service” means little without severity levels and response times. We classify incidents into four levels, each withDefined response and on-site arrival times, with contractual remedies for missed deadlines — a term we proactively include in the contract.
| Tier | Definitions + trigger examples | Response | On-site response | Recovery objective |
|---|---|---|---|---|
| P0 · Emergency | Security incident / regulatory compliance noticeLarge-scale intrusion, ransomware, data leak or on-site regulatory inspection | 5 minutes | 2 hours | 2 hours |
| P1 · Critical | Full failure of core / aggregation / edge equipmentFull-rack network outage, loss of all aggregation-to-core links, or edge-router failure | 15 minutes | 4 hours | 4 hours |
| P2 · Major | Single core / perimeter / WLC failureFailure of any single core switch, edge router, firewall, wireless controller or WAF | 30 minutes | 8 hours | 8 hours |
| P3 · Moderate | Single aggregation switch / departmental access areaAggregation uplink interruption, one VLAN unreachable, or multiple APs disconnecting | 2 hours | Next day | 3 business days |
| P4 · Minor | Single port / user / performance degradationSingle-port disconnect, isolated endpoint slowdown, or sustained high device CPU / memory | 4 hours | 2 business days | As agreed |
| P5 · Inquiry | Configuration changes / capacity planning / optimization recommendationsPolicy changes, new VLANs, capacity assessment and version-upgrade assessment | 1 business day | By schedule | As agreed |
Service components
- Response window8×5 standard,24×7 Optional
- Local spare partsStock optics, power supplies and fans; critical nodesFull-appliance spare
- Regular inspectionsTwice yearly, including configuration baseline and capacity trends
- Change supportChanges during a maintenance window,Includes a rollback plan
- Service coordinationDedicated service channel + service-desk tickets, with a traceable history
- Manufacturer coordinationRMA, firmware upgrades and complex issuesDirect access to manufacturer TAC
- Incident reviewP1 and higher-severity incidents 5 business daysDeliver root-cause analysis and remediation recommendations
- Quarterly reportTicket volume, SLA attainment, capacity trends and recommendations for next-quarter improvements
Emergency procedure:For P0–P2 incidents, open a dedicated service channel immediately. Remote diagnosis, spare-parts dispatch and on-site engineering proceed in parallel, with a complete audit trail.
Eight common scenarios, each with different delivery priorities
The same rack of equipment has very different acceptance criteria for campus, healthcare and education networks. These are eight common scenarios and their key considerations.
Campus office network
Multiple buildings with integrated wired and wireless networks, many endpoints and mobile workers, and frequent policy changes.
Data-center network
A Spine-Leaf architecture carries mainly east-west traffic and is highly sensitive to latency, packet loss and convergence time.
Multi-branch connectivity
Connect headquarters and branches over a mix of private circuits and internet, with centrally deployed policies and zero-touch branch setup.
Industrial and campus low-voltage systems
Factories, warehouses and outdoor campus locations with wide temperature ranges and complex power and cabling conditions.
High-density wireless coverage
Meeting rooms, showrooms and warehouse scanning areas have many concurrent devices and demanding roaming and interference-control requirements.
Education / campus network
Teaching buildings, dormitories, server rooms and labs coexist, with dense devices and frequent roaming. Apply granular rate limits and access controls by identity and time.
Healthcare / hospital network
Hospital wards, operating rooms, imaging and labs run multiple services. Separate internal and external networks and wireless rounds; PACS/HIS imaging and vital-sign traffic need low latency and high reliability.
Retail / chain stores
Street shops, mall counters and warehouse stores vary widely; POS, video backhaul, electronic shelf labels and traffic analytics run concurrently.
What we do and do not do
We define boundaries not to evade responsibility, but toSo you are not left without support in the middle of the night. The two lists below are carried into the contract appendix. During quoting, we identify the owner of anything outside scope.
We are responsible for IN SCOPE
- Architecture design and equipment selection: layered design, three-tier comparison, BOM and quote
- Supply documentation verification: verify serial numbers, source and warranty start date against actual records
- Racking, cabling and integration testing: rack-unit planning, cable labels, optical-loss testing and configuration
- Acceptance testing: measure throughput, failover, convergence, packet loss and full-load PoE, and provide a report
- Manufacturer coordination: RMA, firmware upgrades and complex issues go directly to manufacturer TAC
- Documentation and training: as-built handover, so your team can take over
What we do not do OUT OF SCOPE
- Civil works, cable trays, power-system changes and fire systems: performed by a properly qualified organization
- User endpoints and business systems: operating systems, applications and database configuration are outside network-delivery scope
- Changes to the existing network without written approval: avoid disrupting your active business
- On-site response promises beyond the service tier: no unrealistic verbal promises such as “we will always arrive within 30 minutes”
- Low entry quote followed by add-on charges: BOM items are listed individually; no undeclared additions later
Send us your requirements
A comparable BOM for the agreed scope
You do not need to define the architecture first. Share site conditions and approximate device locations; we will design the layers and compare options — You can use the same BOM to compare quotes; we do not rely on information asymmetry.
You will receive these four deliverables / DELIVERABLES IN 48H
- 01Three-tier comparison— architecture and investment differences shown side by side
- 02Itemized BOM quote— model, quantity, unit price, warranty and lead time itemized
- 03Delivery schedule— timelines, owners and prerequisites for all eight phases
- 04Risks and prerequisites— items requiring your cooperation and factors that may affect delivery