JNCIA-MistAI cheat sheet

Mist is a cloud-native microservices platform hosted on public cloud; management, AI/ML, and data storage run in the cloud while data plane traffic stays on-premises at the AP or switch.

Key architectural components: Mist cloud (management/AI), Mist APs and switches (data plane), and the Marvis Virtual Network Assistant running in the cloud.

Configuration is pushed from the cloud to devices over outbound HTTPS (TCP 443); no inbound firewall ports are required for AP or switch management.

Mist adoption uses device claim codes: an organization-level claim code for APs and an activation code for switches, entered into the Mist dashboard to bind the device.

The Mist dashboard uses role-based access control (RBAC) with roles such as admin, installer, observer, helpdesk, and NOC, scoped at org or site level.

Organizations contain sites; sites group devices, WLANs, and policies. A device belongs to one site at a time but can be moved between sites.

Mist provides open APIs (REST) and webhooks for integration; the same API powers the dashboard, enabling automation and third-party integrations.

Multi-tenancy: a single Mist org can hold many sites, and MSPs use separate orgs per customer with delegated administration.

Mist AI uses supervised and unsupervised ML on telemetry from APs, switches, and clients to detect anomalies and predict issues before users report them.

SLEs (Service Level Expectations) are the core metric model: each SLE has a classifier (success/failure conditions), a metric (0-100 score), and root-cause classifiers.

Wireless SLEs include coverage, capacity, roaming, AP health, client health, time to connect, throughput, and successful connect.

Wired SLEs include switch health, wired client health, and switch throughput; each has its own classifier thresholds.

Marvis is the AI-driven virtual network assistant that consumes SLE data, events, and telemetry to answer natural-language queries and surface root causes.

Anomaly detection compares live behavior against learned baselines per site, AP, or client, reducing false positives from static thresholds.

Mist's AI is cloud-based; APs stream telemetry to the cloud where ML models run, so no local ML engine is required on the AP.

SLE scores are computed over a time window and can be filtered by site, band, client type, and specific classifiers to isolate problems.

802.11ax (Wi-Fi 6) operates in 2.4, 5, and 6 GHz; 802.11be (Wi-Fi 7) adds 320 MHz channels, 4096-QAM, and multi-link operation.

2.4 GHz uses 20 MHz channels with only three non-overlapping channels (1, 6, 11); 5 GHz offers many 20/40/80/160 MHz channels; 6 GHz offers up to 160 MHz with no legacy interference.

Channel width trade-off: wider channels increase throughput but reduce the number of non-overlapping channels and raise co-channel interference.

Transmit power and channel planning interact: lowering power shrinks coverage cells and enables denser reuse; raising power increases coverage but can increase interference.

RSSI is received signal strength; SNR is signal minus noise floor. Typical targets: RSSI better than -67 dBm and SNR of 25 dB or more for voice and real-time apps.

Band steering encourages dual-band clients to 5 or 6 GHz; it improves capacity by moving clients off congested 2.4 GHz.

MIMO and MU-MIMO use multiple spatial streams; OFDMA in Wi-Fi 6 splits a channel into resource units for many small simultaneous transmissions.

BSS coloring in Wi-Fi 6 tags frames by BSSID color to reduce unnecessary CCA deferral from overlapping basic service sets.

Onboarding flow: power the AP, connect it to a network with DHCP and internet access, then claim it in the Mist dashboard using the org claim code.

An AP must reach the Mist cloud (HTTPS 443) to be managed; if it cannot reach the cloud, it stays unadopted and appears as unclaimed or offline.

AP configuration is site-level: assign the AP to a site, then apply site settings such as WLANs, RF profiles, and device profiles.

Device profiles let you apply per-AP overrides such as radio settings, IP configuration, and NTP/DNS; profiles are assigned per device or by rule.

AP models differ by radio capability (e.g., Wi-Fi 6/6E indoor vs outdoor); the dashboard shows model, MAC, serial, and firmware version per AP.

Firmware upgrades are managed from the cloud; you can schedule upgrades per site or device group and the AP checks in to receive them.

Mist APs support mesh and can be configured as a mesh portal or mesh point; the portal needs a wired uplink.

Zero-touch provisioning: a claimed AP downloads its site configuration automatically on first check-in, so no local CLI setup is needed.

A WLAN in Mist defines SSID, security type, VLAN assignment, and band; it is applied to sites and inherited by APs at those sites.

Security options include Open, WPA2-PSK, WPA3-Personal (SAE), WPA2/WPA3 Enterprise (802.1X with RADIUS), and WPA3-Enterprise.

WPA3-Personal uses SAE (Simultaneous Authentication of Equals) to resist offline dictionary attacks; WPA3-Enterprise adds 192-bit mode.

Enterprise WLANs require a RADIUS server profile; Mist supports RADIUS authentication and accounting, plus CoA for session changes.

VLAN tagging: a WLAN can use a static VLAN, a RADIUS-returned VLAN (dynamic VLAN), or a pool of VLANs for load distribution.

Guest networks commonly use a captive portal with a splash page; Mist supports external portal, Click-through, and voucher-based guest access.

WLAN scheduling and rate limiting can restrict SSID availability by time and cap per-client or per-SSID bandwidth.

Fast roaming (802.11r) and 802.11k/v assist clients in roaming; enable them where voice or latency-sensitive clients roam across APs.

Mist Wired Assurance manages supported switches from the cloud; switches must be claimed with an activation code and assigned to a site.

Switches use a switch template or per-switch configuration for ports, VLANs, and profiles; templates are applied at the site level.

Port profiles define speed, duplex, PoE, VLAN membership, and STP settings; assign profiles to ports to standardize configuration.

Wired SLEs (switch health, wired client health, switch throughput) show whether wired users meet service expectations and point to root causes.

PoE budgeting: the switch reports PoE draw per port and total budget; exceeding the budget can de-power ports, so plan per-AP PoE class.

Mist switches support LLDP and can show topology; the dashboard maps switch-to-AP and switch-to-switch links for troubleshooting.

VLAN and trunk configuration: uplink ports carry multiple VLANs as tagged; access ports carry a single untagged VLAN to endpoints.

Switch firmware and configuration are pushed from the cloud; the dashboard shows config status (in sync vs out of sync) per switch.

Marvis is Mist's AI virtual network assistant; it answers natural-language questions about clients, APs, switches, and SLEs in the dashboard.

Marvis Actions surfaces prioritized, actionable issues (e.g., bad cable, DHCP failure, coverage gap) with a root cause and recommended fix.

Marvis client troubleshooting shows a client's connection timeline: association, authentication, DHCP, DNS, and roaming events.

Marvis uses SLE classifiers and telemetry; when an SLE drops, Marvis identifies the dominant failing classifier for that site or client.

Common Marvis findings: PoE negotiation problems, bad Ethernet cable (CRC errors), DHCP scope exhaustion, DNS failures, and sticky clients.

Marvis can be queried in natural language (e.g., 'why is client X slow?') and returns a diagnosis plus supporting evidence.

Marvis Actions are grouped by severity and category, so operators can triage the highest-impact issues first across sites.

Troubleshooting flow: identify the affected SLE or client, read the Marvis diagnosis, verify the evidence, then apply the recommended change.

The Mist dashboard provides live monitoring: AP/switch status, client counts, SLE scores, and event/alarm streams per site.

Alarms are configurable per org or site; events and alarms can be forwarded via webhooks or integrated with ticketing systems.

Mist supports location services using BLE and Wi-Fi; APs with BLE radios can act as beacons or scanners for asset tracking and engagement.

vBLE (virtual BLE) uses the AP's Wi-Fi radio to provide BLE-like location without a dedicated BLE antenna in supported models.

Location accuracy depends on AP density and calibration; asset visibility and analytics rely on multiple APs hearing the same device.

Mist provides a REST API and webhooks for exporting telemetry, SLE metrics, and events into external monitoring or SIEM tools.

Marvis and SLE dashboards are the primary operations views; use them to trend performance over time and compare sites.

Site-level settings control data retention and reporting; export or integrate if you need long-term history beyond the dashboard window.