BackendSide sFlow Analyzer

Changelog

Full release history and version notes for BackendSide sFlow Analyzer – Enterprise Network Traffic Analysis.

Version 4.0

Latest July 2026

A ground-up rebuild of the collector and analyzer — the same sFlow v5 wire parsing and detection catalogue, now with adaptive traffic baselines, deduplicated alerting with email delivery, a live desktop control panel, and HTTPS by default.

Added

  • Adaptive traffic baselines. A new anomaly detector learns what each agent’s traffic normally looks like — per hour of the week, with an overall baseline as fallback — and scores every minute against it, instead of requiring a threshold to be guessed per network. Alerts state what was observed, what was expected, and by how much they differed. Ships in learning-only mode: it studies live traffic and records what it would have raised, without alerting, until you switch it on under Alert Settings.
  • Alert occurrence counts. The Alert History page shows a Last-seen column and an occurrence count, and each alert carries first-seen, last-seen and the number of occurrences.
  • Live control panel. A rebuilt desktop shell with custom window chrome, light/dark themes, separate collector and web-server views, an agent list with live traffic sparklines, and a native About box.
  • Host Watch and dashboard drill-downs. Live-monitor a chosen set of IP and MAC addresses on their own page, and drill into top talkers and conversations from the dashboard.
  • Email alert configuration for custom SMTP, Gmail and SendGrid, with a save-and-test button, cooldown and digest settings, and per-detector thresholds.
  • HTTPS by default on port 8443 with a self-signed certificate generated on first run; drop your own certificate and key into the data directory’s ssl folder to replace it. HTTP on 8080 remains available.
  • Collector health monitoring. A heartbeat every 15 seconds and an anonymous stats endpoint (queue depth, datagrams, drops, insert timing), plus a version endpoint surfaced in the dashboard, the About page and the login footer — so a headless deployment can be monitored without signing in.

Changed

  • A condition that keeps firing is now one alert, not one per check. Repeats update an occurrence count and a last-seen time on the existing alert instead of adding a new row every poll (which used to push the unread badge into the thousands within a day). If the condition worsens, the alert is raised to the new severity and marked unread again, so escalations are not buried. The grouping window is configurable (default 15 minutes).
  • Baseline spike thresholds are now deviation scores, not multiples of a short-term average, so they carry across networks of different sizes; the defaults keep their previous meaning.
  • Alert lists and retention are ordered by last activity, so something still happening stays at the top.
  • Settings are saved to a per-user location, so the application runs correctly from a read-only install (Microsoft Store package or Program Files).
  • The web interface was rebuilt on a new design system with a dark default theme and a light toggle; fonts are bundled, so the dashboard needs no internet access. The routine first-run “certificate created” dialog no longer appears.

Fixed

  • Much higher sustained ingest throughput. The collector now keeps up under heavy load and no longer drops datagrams when traffic bursts.
  • ARP spoofing, MAC flapping and duplicate-IP detectors now fire — each previously read fields its query didn’t produce, so it never triggered. Severity tiers are now respected, and the baseline-spike detector works.
  • Top-N rankings are no longer arbitrary on quiet networks (everything used to tie at a rounded 0.00), and bandwidth figures auto-scale between Mbps, Kbps and bps instead of showing 0.0 Mbps.
  • Empty-state and null-value handling in the interface no longer overflows its container or renders blank rows, and the About box shows the build date correctly.

Security

  • The first login creates the administrator account — no default credentials ship with the application.
  • Agent registration is hardened against a race under concurrent inserts.

Notes

  • Adaptive baseline detection ships in learning-only mode; its hour-of-week baselines need roughly two weeks to mature, so on networks with a strong day/night rhythm it is worth waiting before enabling alerts.
  • The application does not start automatically when Windows restarts — after a reboot, launch it again to resume collection. Counter samples (sFlow format 2) are parsed but not stored.

Version 3.0

April 2026

Email Alert System

  • Email notifications on threat threshold breach — delivered in real time with severity colour-coded HTML messages
  • Three provider options: standard SMTP with authentication, Gmail App Password, and SendGrid SMTP relay
  • Per-severity cooldown timers (configurable for High and Critical) to prevent alert fatigue during sustained attacks
  • Send Test Email button validates SMTP credentials and connectivity without waiting for a live alert
  • All email settings persist in an INI file — survive database resets and application restarts

Expanded Threat Detection — 22 Rules

  • Flood attack detection: SYN, ICMP, UDP, ACK, RST, HTTP, and IP fragmentation floods
  • Amplification and reflection detection: Memcached, DNS, NTP, SSDP, LDAP UDP, SNMP, and CharGen — thresholds sized by real-world amplification factor
  • ARP attack detection: ARP scanning, ARP spoofing, ARP flooding, and MAC flapping
  • Behavioral anomaly detection: IP entropy analysis, baseline traffic spike detection, and Port 0 anomaly
  • Three severity tiers per rule (Medium / High / Critical) with independent email notification threshold per tier
  • Background polling thread evaluates all rules every 60 seconds without impacting collection performance

In-App Notification Bell

  • Live bell icon in the sidebar across every dashboard page — badge shows unread alert count
  • Bell turns red on Critical alerts, amber on High — with a shake animation on new arrivals
  • Click the bell to open an inline preview panel showing recent alerts with severity, type, agent, and relative timestamp
  • Poll interval auto-reduces from 30 s to 10 s while unread alerts are present
  • Mark individual or all alerts as read directly from the panel

Alert History Page

  • Full paginated alert log with filters for severity, alert type, agent IP, and read/unread status
  • Click any row to expand inline — shows alert metadata, detection detail key-value pairs, and raw JSON
  • Opening an unread row automatically marks it read and refreshes the bell badge
  • Per-page stats strip shows Critical / High / Medium / Info counts for the current result set
  • Mark All Read button clears the entire unread backlog in one action

Alert Settings UI

  • Three-tab settings page: Email Provider, Thresholds, and General — no application restart required after saving
  • Per-detection-type toggle, email notification level selector, and Medium / High / Critical threshold inputs
  • Reset individual rows or all thresholds to factory defaults with one click
  • Configurable poll interval, cooldown periods, digest mode, and alert retention window
  • Passwords stored obfuscated in the INI file — masked as *** in the UI after initial save

Version 1.0

January 2026

Initial Release

  • sFlow v5 datagram receiver — listens on configurable UDP port, processes samples in real time
  • 30+ traffic charts covering bandwidth, protocol distribution, top talkers, and error rates
  • Top talkers view — ranked by bytes sent/received per source and destination IP
  • Protocol breakdown — TCP, UDP, ICMP, and other protocol traffic share over time
  • DDoS detection — automatic alerting on abnormal packet rate spikes per source IP
  • VLAN traffic monitoring — per-VLAN bandwidth and packet count graphs
  • MAC address tracking — device identification by hardware address across the network
  • Interface utilisation graphs — inbound and outbound throughput per monitored interface
  • Flow export log — raw sFlow record export to CSV for offline analysis
  • Browser-based dashboard — all charts and tables accessible from any device on the local network
  • Fully on-premises — no cloud dependency, no data leaves the network
  • Supports Windows Server 2008 R2 and later, Windows 10 and 11

sFlow Analyzer

View Product & Download

Versions