## Introduction
Evaluating last-mile customer premises equipment (CPE) and networking gear requires moving beyond manufacturer specifications to focus on real-world reliability in Kenyan ISP environments. While datasheets provide baseline capabilities, actual performance depends on local conditions, integration workflows, and verifiable operational evidence. This guide outlines a practical approach for Kenyan ISPs to assess equipment using documented capabilities and conditional operational advice, ensuring decisions are grounded in observable outcomes rather than assumptions.
## Prioritize Verifiable Evidence Over Marketing Claims
When evaluating CPE or networking gear, ISPs should require evidence that aligns with their specific deployment workflows. For example, instead of accepting a router’s claimed throughput at face value, verify whether the device can maintain stable connections under actual subscriber load patterns in your network. The CENTRANET MikroTik provisioning guide emphasizes that a "provisioned configuration means the approved settings passed readback checks. It does not replace a real customer test." This means ISPs must connect through a customer port, authenticate a test account (Hotspot or PPPoE), and confirm DNS and internet browsing before declaring readiness. Such steps transform speculative claims into observable proof.
## Validate Configuration Workflows, Not Just Hardware
Equipment evaluation must include how well the device integrates into your existing provisioning and management processes. The CENTRANET platform supports MikroTik discovery, configuration, and recorded device evidence as part of network management. When assessing a router, check whether it allows for clear separation of internet, management, and customer connections—a capability explicitly documented in the provisioning guide. Additionally, verify that the device supports automated assignment of clear customer ports while protecting internet and management paths. These operational traits are more telling than raw specs because they directly impact daily workflow efficiency and error reduction.
## Test Recovery and Rollback Capabilities
Reliability in last-mile networks depends not only on uptime but also on how quickly and safely service can be restored after a change or failure. The provisioning guide notes that after approval, the router creates a recovery backup before applying changes, and the agent applies configuration in bounded groups with acknowledgements. When evaluating gear, confirm whether it supports non-destructive rollback—meaning deleting a platform registration does not factory-reset the physical router or remove prior configuration. This ensures that troubleshooting or reconfiguration does not inadvertently erase critical settings, preserving service continuity.
## Use Heartbeat and Telemetry as Operational Evidence
Ongoing reliability assessment should include monitoring tools that provide timely, actionable data. The CENTRANET features page states that ISPs can "check the timestamp of its authenticated heartbeat or agent contact" to determine if a router is connected. However, it also clarifies that "configuration verification and subscriber traffic are separate evidence, so review those as well before declaring the service ready." This means heartbeat alone is insufficient; ISPs must triangulate device contact, configuration readback, and actual subscriber journey tests (e.g., authentication, DNS resolution, browsing) to form a complete picture of link health.
## Assess Flexibility for Mixed Service Types
Many Kenyan ISPs run both Hotspot and PPPoE services simultaneously, requiring CPE that can handle both without compromise. The provisioning guide confirms that "Clear customer ports can serve both Hotspot and PPPoE" and that "the assignment view supports both services on a customer port." When evaluating gear, test whether the device allows concurrent service types on shared ports while keeping internet and management paths protected. This flexibility reduces hardware complexity and supports scalable service offerings without requiring separate equipment per service type.
## Confirm Local Testability Before Scale Deployment
Before deploying any CPE or networking gear across multiple sites, validate performance in a controlled, local environment. The MikroTik guide advises: "Start on a designated test device before changing production traffic." This principle applies broadly—evaluate new equipment using a lab or pilot setup that mirrors your real subscriber workflow, including authentication, billing triggers, and traffic patterns. Only after verifying consistent behavior in this test environment should consideration be given to wider rollout. This minimizes risk and ensures that field issues are identified early.
## Document and Review Configuration Drift
Over time, even reliable equipment may experience configuration drift due to manual changes, firmware updates, or failed provisioning attempts. The provisioning guide recommends that if setup pauses, ISPs should "inspect the reported resource and latest device activity" and use the "Check repaired step and continue" function for a read-only check against approved settings. When evaluating gear, prioritize devices that make configuration state visible and reversible, enabling quick diagnosis without guesswork. This capability supports long-term reliability by reducing mean time to resolve (MTTR) issues.
## Conclusion
For Kenyan ISPs, evaluating last-mile CPE and networking gear should center on verifiable, operational evidence—such as real subscriber journey tests, heartbeat timestamps, configuration readback, and recovery workflows—rather than relying solely on spec sheets. By focusing on what the device demonstrably does in your environment, including how it integrates with provisioning, protects critical ports, supports mixed services, and enables safe recovery, ISPs can make informed choices that enhance network stability and subscriber experience. All recommendations remain conditional on local validation; no equipment should be deemed reliable without testing through your actual service workflow.
## Conduct Real Subscriber Journey Tests with Controlled Variables
To move beyond theoretical performance, ISPs should design repeatable subscriber journey tests that mirror actual service activation and usage patterns. Begin by creating a test account with a known service tier—such as a 10 Mbps Hotspot package—and authenticate it through the customer portal using standard credentials. Once authenticated, verify DNS resolution by querying both local and public resolvers (e.g., 8.8.8.8 and your internal DNS) to confirm proper routing and NAT behavior. Then, initiate a sustained HTTP or HTTPS download test using a tool like wget or curl to a neutral external server, monitoring for throughput stability over a 10-minute window. Avoid relying on speedtest.net-style bursts; instead, look for consistent delivery within 80–90% of the provisioned rate under TCP congestion control. Document any retransmits, stalls, or DNS failures, as these reveal issues not visible in link-layer metrics. This test should be performed on a CPE device connected to a live port in your lab, with the CENTRANET agent active and provisioning completed. Only when the device maintains stable connectivity, correct DNS forwarding, and uninterrupted browsing across multiple test cycles should it be considered for field trial.
## Implement Configuration Drift Detection with Scheduled Readback Checks
Even after successful deployment, configuration drift can silently degrade service quality. To detect this, ISPs should schedule automated configuration readback checks via the CENTRANET platform at regular intervals—such as every 6 hours—and compare the live device state against the approved baseline. Use the platform’s "recorded device evidence" feature to pull the current configuration from the MikroTik router and compare it to the last known good version stored in the system. Focus on critical parameters: NAT rules, firewall filters, DHCP server settings, PPP profiles, and Hotspot walled-garden configurations. If discrepancies are found, trigger an alert and use the "Check repaired step and continue" function to perform a read-only validation without applying changes. This allows technicians to inspect whether the drift resulted from a failed provisioning attempt, local manual adjustment, or firmware-initiated reset. Prioritize gear that makes this comparison transparent and non-intrusive, as devices that obscure configuration state or require CLI-only access increase MTTR during troubleshooting. The goal is not to prevent all changes, but to ensure any deviation is visible, auditable, and reversible through approved workflows.
## Validate Concurrent Service Isolation Under Load
When assessing CPE for mixed Hotspot and PPPoE deployments, ISPs must verify that service types remain isolated under realistic load, not just that they can coexist on paper. Set up a test scenario where 15 PPPoE sessions are active at 5 Mbps each (simulating home users) while 20 Hotspot users authenticate and browse concurrently at 2 Mbps each. Monitor CPU, memory, and queue utilization on the MikroTik device using the CENTRANET telemetry feed or local RouterOS tools. Check for signs of cross-service interference: Hotspot users experiencing PPPoE-style authentication prompts, or PPPoE links dropping during Hotspot portal updates. Verify that firewall mangle rules correctly mark traffic per service type and that queue trees prevent one service from starving the other. The CENTRANET assignment view should show clear separation in the port usage logs, with no overlap in customer-port assignments. Only when both services maintain stable latency (<50ms for gaming, <150ms for video) and zero authentication bleed-through should the device be deemed suitable for mixed-service deployment. This test exposes weaknesses in routing, NAT, or queue management that spec sheets never reveal.
## Establish Handover Records with Signed-Off Test Artifacts
Before handing over a CPE device to field teams or marking it as approved for bulk procurement, compile a handover record that includes signed-off evidence from the validation process. This record should contain: (1) a timestamped screenshot of the CENTRANET verification screen showing "real subscriber journey" completion, (2) a log of the sustained throughput test with min/avg/max values, (3) a configuration diff report confirming no drift from baseline after 24 hours of operation, and (4) a note on recovery behavior—such as whether deleting the CENTRANET registration preserved the prior PPP profiles and Hotspot walled-garden settings. Attach the original test account credentials used (with a plan to rotate them post-handover) and the exact RouterOS version tested. Store this record in a centralized repository linked to the device’s MAC address or serial number in your inventory system. This transforms informal testing into auditable proof, supports warranty discussions, and ensures consistency across engineering teams. Devices that lack clear telemetry, configuration readback, or recovery logs should be scored lower in evaluations, as they increase operational risk despite strong spec-sheet performance.
Discuss your next step
Contact CENTRANET to discuss your network requirements before choosing equipment or changing a live service.