RadixOLT is multi-vendor OLT and ONU monitoring and management software, built for Indian ISPs. It polls every OLT you run, reads optical power on every ONU, and names the port going dark before the phone rings.
Full access for 24 hours. No card. Five short fields and a code from your inbox.
olt-hyd-07Huawei GPON
7 ONUs dark on 1/1/6 — one fibre event, one alert, before the first phone call.Last sweep: all 31 ONUs answering.
OLT · POLLED
olt-bom-01 · ZTE C610
1,024 ONUs · 98.4% working
RX DEGRADING
gpon_olt-1/2/3 · −26.8 dBm
Drifted 3.1 dB in 14 days
LOST LIGHT ×7
one port, one fibre cut
7 ONUs, 1 ticket - not 7
NEXT SWEEP
in 04:12 · every 15 min
Direct SSH · no agent needed
✓ olt-bom-01 · show gpon onu state · 1024 rows parsed! olt-pun-04 · 3 ONUs crossed −25 dBm✓ olt-ngp-02 · display ont info summary 0/0 · 512 rows× olt-hyd-07 · 1/1/6 lost light · 7 ONUs, one alert✓ olt-jai-03 · show ont-find list interface gpon all· olt-sur-06 · auto-poll off · last read 3h ago
✓ olt-bom-01 · show gpon onu state · 1024 rows parsed! olt-pun-04 · 3 ONUs crossed −25 dBm✓ olt-ngp-02 · display ont info summary 0/0 · 512 rows× olt-hyd-07 · 1/1/6 lost light · 7 ONUs, one alert✓ olt-jai-03 · show ont-find list interface gpon all· olt-sur-06 · auto-poll off · last read 3h ago
Six terminal windows and a WhatsApp group.
That is how most fibre plants are actually run. Every vendor speaks its own CLI. Optical power lives in a command nobody remembers. A fibre cut arrives as forty calls to the helpdesk instead of one alert on one port.
The information exists. Your OLTs report it every time you ask. Nobody is asking on a schedule, and nobody is watching the trend.
×Truck rolls dispatched to guess at the fault
×One engineer who knows the ZTE syntax, on leave
×Churn you attribute to price, not to signal
4 TABS OPENputty · 10.44.2.11 · telnet
ZXAN# show pon power onu-rx gpon_olt-1/2/3
ONU-Index Rx-Power(dBm)
1/2/3:1 -21.44
1/2/3:2 -26.81
1/2/3:3 -27.93
1/2/3:4 -22.07
ZXAN# show gpon onu state | include offline
1/2/3:7 offline
1/2/3:8 offline
1/2/3:9 offline
-- was that yesterday's reading or today's? ZXAN# _▌
Four screens run the whole plant.
The product's own screens, filled with sample data.
Screenshot preview of the RadixOLT dashboard, alerts and terminal screens, following the captions beside it.
app.radixolt.com
AM
DashboardAlertsTerminal
Search ISPs, OLTs, ONUs, jobs…⌘K
Dashboard
NorthArc Fiber
Fleet as of 2 minutes ago·34 of 41 OLTs auto-polling
Poll all
Refresh
Needs attention now
6critical alerts unacknowledged3OLTs stale or never polled
FLEET AT A GLANCE
FLEETOLTs →
41OLTs
Active36
Inactive5
Auto-poll34
38 of 41 polled at least once
ONUSONUs →
38,41297.6% working
Working37,489
Offline612
Dying gasp217
From 38 OLTs · 3 never polled
OPEN ALERTSAlerts →
48open
Critical14
Warning22
Unacked19
6 of 14 critical still unacknowledged
OPTICAL HEALTH
Signal health
ONU RX power, fleet-wide · 31,204 reporting, 7,208 with no reading
−32 dBm−28 critical−25 degrading−8 dBm
1,284
critical ≤ −28
3,061
degrading ≤ −25
26,859
healthy
Light lost on one port
2 or more ONUs losing light on one PON port - one fibre event, not many
7
olt-hyd-07 · 1/1/6
4 minutes ago
4
olt-pun-04 · 0/2/1
22 minutes ago
3
olt-bom-01 · 1/2/3
1 hour ago
2
olt-jai-03 · g0/4
3 hours ago
16 dark ONUs, 4 fibre events to dispatch
Alerts
48 open · 19 unacknowledged · 29 being worked · 7 resolved today·as of 2m ago
STILL FAILING
31
SAFE TO CLOSE
12
NOT READ
5
NEW · LAST HOUR
9
Search customer, MAC or OLT…
KindOLTSeverityNewest first48 rows
FAULTSUBSCRIBERWHYCURRENT STATELOCATIONOPEN FOR
OLT downEvery ONU on this OLTSNMP 10.20.4.15:161No SNMP replylast reply 1h agoStill failingolt-ind-051h
Offlinekiran.sHWTC6F21A0C3LOSStill failingolt-hyd-071/1/6 · ONU 34mAcknowledge
Offlinepriya.mHWTC1B88E2D4LOSStill failingolt-hyd-071/1/6 · ONU 54m
Low opticalsunita.rZTEGC4A2215E−28.6 dBmStill failingAcknowledged 12m agoolt-bom-011/2/3 · ONU 818m
Offlineanil.kHWTC9A0F31B2LOSiRecoveredolt-ngp-020/1/4 · ONU 412h
OLT rebootedEvery ONU on this OLTSNMP 10.20.7.2:161Uptime resetbooted 3h agoRecoveredolt-sur-063h
PICK AN OLT
olt-bom-01
ZTE GPON · Direct SSH
olt-hyd-07
HUAWEI GPON · Direct SSH
olt-jai-03
GENEXIS GPON · Direct SSH
olt-pun-04
TPLINK GPON · telnet
ZTE QUICK COMMANDS
show gpon onu stateshow cardshow pon power onu-rx …show optical-module-info …show snmp
CONNECTEDolt-bom-01 · via Direct SSH · session 00:04:18Disconnect
ZXAN# show pon power onu-rx gpon_olt-1/2/3
ONU-Index Rx-Power(dBm) Status
1/2/3:1 -21.44 working
1/2/3:2 -26.81 working
1/2/3:3 -27.93 working
1/2/3:4 -22.07 working
1/2/3:6 -- dying-gasp ZXAN# show gpon onu detail-info gpon_onu-1/2/3:3
ONU interface : gpon_onu_1/2/3:3
Name : ZTEG-C8A41F2D
Phase state : working
Online duration : 14 days 06:22:41
Last down cause : LOS
ZXAN# pon-onu-mng gpon_onu-1/2/3:3▌
Destructive command - pon-onu-mng
This runs live on olt-bom-01 (10.44.2.11) and takes effect immediately. Terminal sessions leave no record, so there will be nothing to review afterwards.
Run anywayCancel
Reach it. Ask it. Compare it. Fix it.
No agents on your ONUs, no changes to your network design, no new protocol. Just the reads your OLTs already answer, on a schedule, compared against the same thresholds everywhere.
01CONNECT
Reach the chassis, however it's routed.
Direct SSH when the OLT has a route from your cloud. When it doesn't, run RadixOLT inside your NOC instead - same reads, same screens. A Site Agent that dials out from the NOC LAN is planned. SSH or Telnet, per-OLT ports, credentials entered once.
executordirect SSH · site agent planned
protocolSSH 22 · TELNET 23
snmp161 · optional, per OLT
secretnever returned by any API
02POLL
Ask in the vendor's own language.
One auto-poll job per OLT, every 15 minutes by default. Each platform's adapter runs that platform's real command - the one your engineer types - and normalises every row into a single shape, so nothing downstream ever branches on vendor.
▸auto-poll · olt-bom-01 · get_onu_state
adapter ZTE (ZXA10 C610)
command show gpon onu state
status SUCCESS · 1,024 ONU rows parsed
next run in 11m 04s
03DETECT
One set of thresholds, fleet-wide.
A reading means the same thing on every screen and every vendor. Line faults get grouped by PON port, because two on one port is one fibre event and not two tickets. And if an OLT hasn't answered in six hours, the dashboard says you are blind on it rather than showing you a stale green.
RX ≤ −28.0 dBmcritical
RX ≤ −25.0 dBmdegrading
≥ 2 gasps · one portcluster
no reading > 6 hstale · flagged
04ACT
Tell someone. Then fix it from the same screen.
Alerts carry a severity and an acknowledgement, and route to Telegram so the on-call sees them without opening a browser. WhatsApp is next on the roadmap. From the alert, open a live session on that OLT with its own commands one click away - and a hard stop in front of the destructive ones.
CRITICALolt-hyd-07 · 1/1/6 · 7 gasps
Telegram to NOC on-call, sent 4 min ago
Session opened by Priya N. · 00:04:18
The alert leaves the building. On Telegram.
Nobody watches a dashboard at 03:00. The sweep raises the alert, and the same alert goes to the on-call's phone in the app they already have open - with the OLT, the PON port and the count in the first line, so a groggy engineer knows whether to get up. What is sent, and when.
WHAT IS SENT
LINE FAULTTelegram, the moment the sweep that read it ends
LOW OPTICALTelegram, with the reading, at −28 dBm or weaker
OLT DOWNTelegram after two missed checks, and again when it answers
01Past ten at once, one message per PON port - a cut fibre is one buzz, not fifty
02Planned: quiet hours 23:00-07:00 hold the warnings. Criticals ignore them
03Planned: every send stores a delivery receipt on the alert - you can prove it went out
04Contact groups per ISP or per OLT - a franchise never sees another's plant
Bot token and sender number sit behind the ISP Admin role. They are secrets, so no API returns them.
BOT ONLINE
CRITICALRadixOLT· NOC on-call
olt-hyd-07 · PON 1/1/6
7 ONUs lost light inside 40 s. Grouped as one fibre event - check the feeder before the drops.
21:14·Open in apppreview, not live yet
WARNING15-min digest
olt-bom-01 · 1/2/3:3
RX −28.6 dBm, past the −28 line on a working ONU - splitter or a bend, not the customer's router.
21:16preview, not live yet
RESOLVEDclosed by Priya N.
olt-hyd-07 · PON 1/1/6
All 7 ONUs back online. 41 minutes down, splice replaced at the chamber.
21:55preview, not live yet
CRITICAL+91 ⟨on-call⟩
olt-hyd-07 · PON 1/1/6 · 7 ONUs
Light lost on one port, one fibre event. Reply ACK to acknowledge without opening the app.
21:14·template alert_critical_v2read 21:14
ACK
21:15 · Priya N.alert acknowledged in app
DIGEST21:30 sweep
olt-pun-04 · 3 ONUs crossed −25 dBm olt-ind-05 · no reading in 9 h · auto-poll off olt-ngp-02 · 1 ONU offline 31 min
21:30read 21:31
Is it the fibre, or the bill? One screen now says which.
Link the RADIUS server your subscribers already log in through - OneRADIUS or Xonware - and each subscriber lands on the ONU their router sits behind, with their plan, expiry and live session beside the optical reading. A dark ONU on a lapsed plan needs a renewal call, not a truck. How the link works.
HOW THE LINK IS MADE
MACEach OLT is read for the router MACs it has learned behind every ONU, in one pass of the chassis
MATCHEvery MAC is matched to the login your RADIUS server recorded it on
LINKThe subscriber is placed on that ONU, and moves with the router when it moves
01OneRADIUS brings the whole subscriber roster on the first sync.
02Xonware subscribers appear as their routers are seen behind an ONU.
03Your RADIUS server stays the record: names, mobiles and addresses are read from it, never typed in twice.
04A missed read unlinks nobody. An absent MAC is not proof a subscriber has gone.
The MAC read covers Genexis, DBC, Syrotech and Optilink chassis today. On other platforms the sync names the OLTs it could not read, rather than skipping them.
ONU 1/1/6:15
olt-hyd-07 · last seen 23 Sep, 22:14
OFFLINE
LINKED SUBSCRIBERXONWARE
Rahul G.
rahul.g
PLAN
100 Mbps Home
EXPIRY
Expired 23 Sep
SESSION
None
The plan lapsed the evening the ONU went dark. This is a renewal call, not a fibre fault.
Link Radius Server is a premium add-on, switched on per ISP by our team. It is not part of the 24-hour trial. See how it is priced.
Nine platforms, confirmed against live chassis.
Not a generic SNMP walk with vendor logos pasted on top. Each adapter runs commands we have actually executed on that platform - and where a vendor's CLI splits its own product lines, we treat them as two platforms, because they answer nothing alike. Compare every platform.
Inventory them, attach credentials, open a terminal - all of that works. Automated reads do not, and we will tell you that before you sign, not after. Ask us where they sit on the roadmap.
Run six ISPs from one login. Or lock a viewer to two screens.
Four roles, and every one of them is enforced on the server - not hidden in the navigation. A Super Admin switches between ISPs from the sidebar; an ISP Admin never sees that the other ISPs exist. Your franchise partners get their own tenant and no view of each other's plant.
Terminal access stops at Operator. Credentials and contacts stop at ISP Admin - a Telegram bot token is a secret, so it lives with the admin screens.
You are handing us the keys to your plant. Here is exactly what we do with them.
CREDENTIALS
Secrets never come back out
No endpoint in the product returns a stored password or SNMP community - the single-OLT fetch returns the username and nothing else. You cannot exfiltrate what the API will not say.
ACCESS
Roles checked twice
The sidebar hides what your role cannot use, and the server refuses it independently. Two-factor sign-in is planned.
GUARD RAILS
Destructive commands stop and ask
A command that can drop a customer needs a second, explicit confirmation naming the OLT and its IP - and it tells you plainly that a terminal session leaves no record to review afterwards.
DEPLOYMENT
Our cloud, or entirely in your NOC
Containerised and self-hostable behind your own nginx. If management IPs must never leave the building, run the whole thing inside it and keep the data at home.
REACHABILITY · PLANNED
No inbound holes
A Site Agent that dials out from your LAN, so an OLT with no public route stays that way. Until it ships, an OLT with no public route needs the self-hosted install.
HONESTY
Stale data announces itself
A port whose link state the platform cannot read shows as “not reported”, never as up. Six hours without a reading is shown as blindness, not as health. Wrong-but-green is worse than nothing.
Start on the 24-hour trial with every plan feature switched on, then pick the tier you actually need. Nothing you set up during the trial is lost when it ends - it waits for you.
Basic
Run the fleet. See the faults.
₹599per OLT / month
Multi-OLT management across every supported vendor
Scheduled polling of every ONU, every 15 minutes
Optical power, ONU state and PON port monitoring
Standard in-app alerts for line faults and low RX
Guard-railed terminal on the chassis
Roles and per-account OLT scoping
Recommended
Pro
The alert leaves the building.
₹799per OLT / month
Everything in Basic, plus:
Telegram alert delivery to your on-call groups
PlannedWhatsApp alert delivery
Advanced customer fibre mapping on the plant map
Fibre-break location down the PON tree
PlannedField and wall-display map views
Link Radius Server
Premium add-on
Subscriber names, plans, expiry and live sessions from OneRADIUS or Xonware, placed on the ONU each one sits behind. Switched on per ISP alongside either plan, and priced with you on the call.
Card payments are not live yet, so choosing a plan leaves your details and we call you back within one working day to set it up. A trial covers up to 5 OLTs and polls from every 15 minutes; both plans lift that. Start the 24-hour trial if you have not already, or read the plan details and refund terms.
talk to us
Bring two of your own OLTs to the call.
Huawei, ZTE, Genexis, TP-Link, Syrotech, DBC, Optilink or EasyPath - give us reachability and a read-only credential and we will poll them live while you watch. Forty minutes, your plant, no slides.
✓Read-only credential is enough for the demo
✓Self-hosted install if the OLT has no public route