Backup Watch: the alert nobody else gives you

Server monitoring and backup visibility.
Unified. In one place.

Sergward brings together the health of your Windows, Linux and Mac servers, the network gear that will never run an agent, and your websites — with the real state of every backup in the same panel. And it tells you once, via Telegram, Slack, Teams, email or a webhook.

Start free See how it works

one binary · no open ports · free for 5 hosts · no credit card

app.sergward.comHostsBackupsAlertsDevices

Host	OS	CPU	RAM	Disk	Backup	Status
SRV-CONTA-01	Win 2022	12%	64%	71%	4 h ago	Up to date
SRV-SQL-02	Win 2019	34%	82%	88%	26 h ago	Backup failed
web-prod-01	Ubuntu 24	8%	41%	52%	1 h ago	Up to date
SRV-ARCHIVOS	Win 2016	6%	38%	93%	6 h ago	Disk 93%
SRV-SUCURSAL-N	Win 2019	—	—	—	—	No signal 07:00

Sergward via Telegram03:12

SRV-SQL-02 — Backup failed. Job "SQL-Nightly" has had no successful run for 26 hours.

View details · Create ticket · Snooze 1 h

01The problem

The backup nobody watches

Everyone monitors CPU and disk. Almost nobody finds out the backup has been failing for six days — until they need it. And then no alert can save you.

02The tools

Broken at both ends

The classics (PRTG, Zabbix, Nagios) are powerful but take forever to configure. The modern ones (Datadog, New Relic) cost a fortune. And none of them tells you whether the backup ran.

03The answer

A single guardian

Server health, the switch that will never run an agent, the certificate about to expire and last night's backup — in one panel, with one source of alerts.

Backup Watch

Did last night's backup run?

The question that defines your week, answered every day. Sergward reads the real result of every job — Veeam, Windows Server Backup, restic, borg, rsync, cron — and alerts “X hours without a successful backup” per job, each with its own threshold.

Catches the job that failed and the one that never ran at all
Unified view: the backup right next to server health
An old age says nothing about right now: if the host stopped reporting, the state turns unknown instead of staying green

Try it with my backups See supported engines

Veeam · Daily-Full4 h ago · 218 GBup to date

Windows Server Backup · SRV-FILES6 h ago · 54 GBup to date

Veeam · SQL-Nightly26 h without a successful runfailed

restic · /var/www1 h ago · 2.1 GBup to date

cron · nightly dump3 h ago · 800 MBup to date

Architecture

An agent that pushes.
A platform that thinks.

You install one binary per server. It collects metrics, services, events and the result of every backup — and along the way it probes the network gear around it, which will never run an agent of its own. Everything leaves over outbound HTTPS/443, without opening a single port. The platform receives it, stores it, evaluates the rules and fires the alerts.

Collects

System metrics

Services and processes

Event Log and syslog

Backup jobs

Network gear over SNMP

Websites and certificates

push https/443

Alert enginerules · dead-man switch
dedup · snooze · escalation

notifies

Acts

Telegram, Slack, Teams

Webhooks and syslog

A ticket in Sergward

Dashboards and reports

The platform

Everything it watches, for real.

Not a list of promises: this is what the product does today, piece by piece.

Core monitoring

The health of every host, live

CPU, RAM, disks, network, uptime, services and processes. Plus the SMART health of every drive: temperature, life left and power-on hours, which is what warns you before it dies. Sane thresholds out of the box: install it and it already alerts well.

CPU

12%

RAM

64%

C:\

88%

SQL Server (service)

running

Alert engine

Alerts that don't add noise

A dead-man switch: it tells you when a host stops reporting, not when it reports something bad. It is the one outage no metric can ever report, because whatever went down stopped sending anything. Plus deduplication, snoozing through maintenance windows and escalation when nobody answers.

03:12 SRV-BRANCH-N no signal — Telegram to on-call

03:19 no answer for 7 min — escalates to email + IT lead

03:24 acknowledged by marcos — snoozed 1 h

03:41 host reporting again — resolved

Gear without an agent

The switch, the printer and the UPS

Scan a range from any host and the gear worth watching shows up. It is monitored over SNMP and ping, with nothing installed on it: interfaces, temperature, battery, toner. And it receives traps.

SNMP sw-core-01 — 24 interfaces, 2 down since yesterday

SNMP ups-rack-a — 40 min of runtime, on battery

ping printer-acct — 8.8 ms, no loss

Sites and certificates

The certificate that expires on a Sunday

HTTP and HTTPS checks from the platform: response code, latency, a keyword that must appear on the page, and the days left on the certificate. The one that expires without warning is the one that takes your site down over a long weekend.

200 www.client.com — 240 ms · certificate: 68 days

200 portal.client.com — certificate: 9 days

503 api.client.com — no answer for 4 min

Dashboards and reports

The screen your team looks at

Build dashboards from the widgets you want, pointed at one host, at a tag or at the whole fleet. TV mode for the office monitor, templates so you don't start from scratch, and a report that goes out by email with your logo on top.

▦ Fleet status — 7 widgets · TV mode

▦ Disk space, by machine — by tag

✉ monthly report — sent to 3 recipients

One organization per client

For MSPs: every client, separate

An isolated organization per client: their hosts, their devices, their alerts, their dashboards and their users never mix with anyone else's. With roles, two-factor you can require across the team, and your logo in the portal and on the reports.

MSP Norte ITyour organization

Clínica San José18 hosts · up to date

Estudio Contable RV6 hosts · 1 alert

Distribuidora Andes31 hosts · up to date

What it watches

All of this is in the free plan.

No premium modules and no sensors billed on the side. What you see is what it does, and the agent detects on its own what runs on each machine to suggest what fits.

What	What it covers	System	How it gets there
Servers and workstations	CPU, RAM, disks, network, uptime, services, processes and a dead-man switch	Win / Linux / Mac	Agent
Backup Watch	Veeam, Windows Server Backup, restic, borg, rsync and cron jobs	Win / Linux	Agent
Drive health	SMART: temperature, life left, power-on hours, reallocated sectors	Win / Linux	Agent
Inventory	Hardware, installed software, BIOS, serial number and model	Win / Linux / Mac	Agent
Events	Windows Event Log and Linux syslog, filtered: crashes, failing disks, reboots, repeated auth failures	Win / Linux	Agent
Network gear	Switches, routers, firewalls, UPS units, printers and NAS over SNMP, with traps and scan-based discovery	No agent	Probe
Websites	HTTP and HTTPS uptime, response code, latency, keyword and certificate expiry	No agent	Platform
Tickets	From the alert to an assigned ticket, with the host's metrics and events inside it	Any	Platform

Integrations

It fits the stack you already have.

The rule for every integration: Sergward joins your workflow, it doesn't ask you to change it.

Alert channels

The alert lands where your team already works. If a channel fails it retries and falls through to the next one, and every delivery is recorded.

SlackTMicrosoft Teams

EmailWWebhooksssyslog / SIEM

Backup engines

Backup Watch reads the real result of each job from the engine you already run — it never asks you to switch backup tools.

VVeeamWWindows Server Backuprresticbborgrrsync / cron

Operating systems

One binary per system, no dependencies. Genuinely mixed estates, which is what real life looks like.

WWindows

Linux

Apple / Mac

Network gear

Whatever can't run an agent gets probed from one that can, inside the same network.

SSNMP v1 / v2c / v3tTrapsLCDP / LLDPpICMP / ping

Deployment and security

Try it locally. Scale it in the cloud.
And your security team signs off.

The same binary either way — moving between them is one line of configuration, not a migration. And every design decision is made to clear a corporate EDR.

Local

Just the agent, with its embedded dashboard listening on your own machine. To try it in two minutes, with no account and no cloud.

sergward-agent.exe install
# dashboard: 127.0.0.1:8484

Cloud

We run the platform. You paste your organization's token and the host shows up on its own.

sergward-agent install \
  --token SGW-XXXX

Outbound push, port 443

Zero inbound ports, zero firewall rules. The agent only emits — nobody reaches into your server from outside.

Signed binaries

Code signing on every release, no packers and no downloads at runtime. The PowerShell ships signed, on disk and readable.

Metrics only

System counters, service states and job results. Never file contents and never your users' data. The list is public.

Two-factor you can require

Anyone can turn on their own second factor, and the account owner can require it across the organization. In an MSP, the weakest password on the team is the real attack surface.

Comparison

What you're going to compare us against.

And you should compare — this table is honest, including about what we don't have.

Product	Base price	Model	Windows	Spanish	Unified backup	Agentless network
PRTG	~US$179/mo	Per sensor	Yes	No	No	Yes
Datadog	~US$15/host	Per host	Partial	No	No	Yes
Netdata	~US$4.5/node	Per node	Limited	No	No	No
Zabbix / Nagios	Free	Open source	Partial	No	No	Yes
Sergward	US$0 to US$6	Per host	Yes	Yes	Yes	Yes

Public reference pricing for 2025–2026, approximate. Zabbix and Nagios cost nothing to run: the cost is the server and the hours it takes to configure them.

Pricing

Three options. Per host, and that's it.

Count your hosts and you know what you pay. No mystery sensors, no modules billed on the side and no surprises at renewal: everything the product does is in all three plans.

FreeUS$0

Up to 5 hosts, forever
Everything the product does
Backup Watch on all of them
Network gear and websites
Every alert channel
No credit card

Start free

ProUS$6/host/mo

Unlimited hosts
Long history
Reports by email
Read-only API and webhooks
Priority support

Choose Pro

For MSPsMSPUS$4/host/mo

Everything in Pro
One organization per client
Your logo in the portal and reports
Two-factor required across the team
A single invoice

Talk to us

Frequently asked questions

What sysadmins ask us.

Does Sergward take the backups?

No, and that's on purpose. Sergward is the intelligence layer over the backups you already run — Veeam, Windows Server Backup, restic, borg, rsync, cron: it detects whether the backup actually ran and alerts you if it failed, or if it never ran at all. It unifies visibility, not execution.

Do I have to open ports on the firewall?

No. The agent pushes over outbound HTTPS/443, the same port the server already uses to browse. Zero inbound ports, zero firewall rules, nothing to touch on the router.

What exactly does the agent collect?

Only system metrics and states: CPU, RAM, disk and network counters, service and process states, SMART drive health, hardware and software inventory, critical Event Log entries and backup job results. Never file contents and never your users' data. The full list is public and lives on this same site.

How does it monitor a switch or a UPS, which can't run an agent?

One of your hosts acts as a probe: it scans the range you point it at, finds the gear and queries it over SNMP and ping from inside your network. Nothing gets installed on the switch and nothing gets exposed to the internet.

Is the free plan a trial with an expiry date?

No. Five hosts free, forever, with everything the product does: Backup Watch, network gear, websites and every alert channel. We don't ask for a card. If your estate grows, you move to Pro.

How many servers does it handle?

The agent uses under 30 MB of RAM and about 1% of CPU. The platform scales from 2 hosts to thousands; the ceiling we publish comes from a real load run, not an estimate.