Solution · Servers

Windows server monitoring with nothing to install on the servers


The agent rollout is the part of a monitoring project that goes to change control, gets scheduled for the third Thursday, and quietly skips the two servers that matter most. SysPeep reads Windows over WMI, from one service, and skips the rollout.


Key points

  • Windows servers over WMI: services, disks, CPU and memory, Hyper-V memory pressure, DNS server counters, terminal sessions, battery charge, and temperatures and fan speeds where LibreHardwareMonitor is on the host.
  • A WMI explorer: browse a host's classes property by property, with sample values and units where the host reports them, and turn any numeric property into a graphed, thresholded sensor with no script to write.
  • Windows event log subscriptions become alerts through rules with severity mapping; the rest stay on the events timeline.
  • Any SQL result as a sensor; HTTP sensors and multi-step web journeys for IIS and the applications on it; server hardware over IPMI and Redfish while the OS is down.
  • Linux hosts over SSH, vCenter and Proxmox for the hypervisors, Prometheus endpoints scraped rather than duplicated. One console, no agents.

On this page: Over WMI The WMI explorer Event logs Databases and web Hardware and hypervisors The poller Where it runs


Most of what goes wrong on a Windows server announces itself before it fails: a service that stopped, a volume filling, a certificate with three days left, an event log line that nobody subscribed to. SysPeep watches all of it over WMI from the monitoring server, so the only thing that needs installing is the one service you already run, and the only thing that needs opening on the hosts is what WMI already needs.


What it watches over WMI


What SysPeep doesWhat it means on call
Services, disks, CPU, memory, Hyper-V memory pressure, DNS server counters, terminal sessions, battery charge, and more, over WMIDeep Windows visibility with no agent to install, patch or explain to change control on every server
Per-instance sensors: one sensor for a set of volumes or services, whose status is the worst instance's and whose value is the worst, the total, or "2 of 9 failing" as the type demands, each instance its own chart lineOne full drive among nine is reported as the problem, not averaged away
Countdowns as dates: disk space and certificate days are projected to the date they run outA ticket with lead time instead of an incident at 3 a.m.
Down versus degraded: only the sensor you chose can mark the server down; a stopped service or a full disk marks it degraded with its own alertThe availability record stays honest, and the ticket-worthy problems do not page the on-call phone

The counter nobody pre-packaged


The catalog holds 280 sensors across 38 collection methods, and every site still has the counter none of them quite fit: the line-of-business service's own performance class, a queue length, a licence count exposed by a vendor's WMI provider. The WMI explorer lets you browse the host's classes property by property, with sample values and units where the host reports them, and turn any numeric property into a sensor. It is graphed, thresholded and alerted on like a built-in one, and it does not add to the license count.


Beyond WMI, the output of a command, script or program becomes a sensor, as does a file or folder check (the backup job that proves it ran by writing to a drop-folder) and a passive push from a batch job that cannot be polled. Every sensor can be tested against the server before it is added, and a card names which credential the test will use.

Beyond the catalog

  • Any numeric WMI property, one click
  • Command, script or program output
  • File and folder checks
  • Any SQL result
  • Pushed readings from jobs
  • Calculated and composite sensors
The script you already run by hand on Monday mornings becomes a line on a chart with a warning level attached.

Event logs: the one that matters, and the thousand that do not


Windows event log subscriptions arrive at the same service as syslog and SNMP traps. Rules turn them into alerts by kind, application and severity, with a built-in tester that shows which rule wins for a given event. The event that matters reaches the phone; the thousand that do not stay on the events timeline beside the server's graphs, where they are still there when you need to know what happened right before something broke. The syslog and trap receiver page covers the event side in full.


Databases, IIS and the applications on the box


  • Database query sensors: any SQL result as a sensor, so replication lag, order backlog and queue depth are monitored beside the server they run on; native health for Redis, MongoDB and RabbitMQ.
  • HTTP sensors for IIS and anything behind it: status, latency, JSON and text matching, certificate days remaining, minimum TLS version, weak ciphers, security headers.
  • Multi-step web journeys: log in, navigate, verify, so "the site is up but nobody can log in" is caught as the outage it is.
  • Mail flow round trips, LDAP bind and RADIUS checks for the Exchange, directory and access servers that lock everyone out at 8:55 a.m. when they fail.

Hardware, hypervisors and the Linux box in the corner


While the OS is down

IPMI and Redfish reach the server's management port for BMC temperatures, fans and power, so a hardware fault is seen before the operating system notices, and while the operating system is down. Fans and most temperatures on the host itself come through LibreHardwareMonitor where it is installed.

The rest of the estate, same console

VMware vCenter, Proxmox and Kubernetes for the hypervisor and container estates; Prometheus and OpenMetrics endpoints scraped rather than duplicated; Linux hosts over SSH, where whatever the command prints is the reading. One alert model and one on-call process for all of it.


A poller that does not let one slow server cost the rest a sample


All of a server's WMI queries are in flight at once, and sixteen collection workers by default (adjustable from 1 to 64) keep one session per device and method, so one slow server no longer delays everyone else's readings. Polling runs on a fixed 30-second clock grid with a bounded cycle; a hung host delays only its own sensors, and a failed WMI check raises an alert that reads "error:" and the reason, not a flat line. When a poll takes longer than its interval, the chart says so in plain words with the advice to lengthen it.


Where it runs, and what it needs


SysPeep installs from one MSI as a single Windows service on a Windows server or desktop you already have, with the console compiled into the service executable, time-series data in its own embedded store, and configuration and history in SQLite, MySQL or SQL Server. Device credentials are encrypted with AES-256-GCM under a key kept in the installation's own data folder. Quick Start adds the firewall rules the service needs. The deployment page has the details; the agentless and extensible page covers every collection method.


Questions administrators ask about Windows monitoring


Does it monitor Windows 10 and Windows 11 desktops as well as Windows Server?

Yes. Any Windows machine reachable over WMI is a monitored device: Windows Server 2012 and later, Windows 10 and Windows 11. The server component itself also installs on a Windows desktop, so a net monitor for Windows 10 or Windows 11 can be the same machine that monitors the network.


Can it monitor SQL Server performance and replication (MS SQL, MSSQL)?

SQL Server exposes its performance counters through WMI, so batch requests, buffer cache hit ratio, blocked processes and the rest are one click away in the WMI explorer, and a database query sensor turns any T-SQL result into a reading: replication lag, queue depth, order backlog. Monitoring SQL Server performance beside the host's CPU, memory and disks is what the device page is for.


Can it monitor an Apache web server, IIS or nginx?

HTTP sensors check status, latency, content, certificate days remaining and TLS settings against any web server, and multi-step web journeys log in and verify the pages behind it, so an Apache web server monitor and an IIS monitor are the same sensor pointed at different ports. Apache on Linux is reached over SSH for its process and log checks as well.


Does it monitor Windows services, and can it restart them?

A Windows service sensor watches the state of a service or a set of services, with a per-instance view that reports "2 of 9 stopped" rather than an average, and raises a degraded alert when one stops. Restarting is a job for your automation; a script notification can trigger it, with the audit log recording who configured that.


What Windows network diagnostics does it run: TCP port monitor, network connection monitor, internet connection monitor?

TCP port sensors prove a service is listening; ICMP sensors prove a host answers; DNS and NTP sensors prove the resolvers and clocks; an HTTP sensor against a public address is an internet connection monitor with a latency graph. All of them run continuously from the Windows server, on a 30-second clock, so the intermittent fault is on the graph when you look.


Is there a server monitor screen or a wall view for the NOC?

The NOC wall view is designed to be read from across the room: fleet status, open incidents and the devices that need attention, in any browser. The dashboard and the device workspace are the screens for the person at the desk.


Does it monitor Dell and HP server hardware?

Through IPMI and Redfish on the management port (iDRAC, iLO and their equivalents): temperatures, fans, power and hardware health, available while the operating system is down. On the host itself, fans and most temperatures come through LibreHardwareMonitor where it is installed.


Can it monitor Windows Server hardware usage: CPU, memory, disk and disk I/O?

CPU, memory, per-volume disk space with the date each volume will fill, disk queue and I/O counters, network interface counters and the process list, all over WMI with no agent. Per-instance sensors mean one full drive among nine is reported as the problem, not averaged away.


Does it track internet usage on Windows 10 machines?

A host's own network interface counters are available over WMI, so the traffic each Windows machine sends and receives is a sensor. For what that traffic was, the flow records from the router or switch show the conversations; SysPeep does not inspect what users browse.


Is it Windows based network monitoring software, and what does the server need?

Yes: one MSI, one Windows service, on a Windows server or desktop you already have, with the console compiled in and time-series data in an embedded store. Configuration and history live in SQLite by default, or MySQL or SQL Server when policy requires. Quick Start adds the firewall rules.


What are the best server monitoring tools for Windows, and how does SysPeep compare?

The category splits into agent-based suites, cloud services billed per host, and agentless on-premises tools. SysPeep is in the third group: nothing installed on the servers, data on your own server, licensed per device with unlimited sensors on a perpetual license, and free for 5 devices. Which is best depends on whether you want to install agents and whether you want to own the tool.


Can it monitor file permissions or permission changes on a Windows server?

Not as a permissions auditor. What it does is subscribe to the Windows security event log, so an audit event for a permission change becomes an alert through a rule, and the event stays on the timeline with the rest.


Does it monitor Hyper-V hosts and virtual machines?

Hyper-V memory pressure and the host's counters come over WMI; the guests are monitored as devices of their own over WMI or SSH; VMware vCenter and Proxmox estates are read through their own APIs. One console, one alert model, for the hypervisor and what runs on it.


What about Exchange, Active Directory and domain controllers?

Mail flow round trips through SMTP, IMAP and POP3 catch a stuck Exchange queue minutes before the complaints; LDAP bind and RADIUS checks prove the directory and network-access authentication continuously; DNS server counters and the domain controller's services and event logs come over WMI.


Does it monitor Linux servers and their processes too?

Over SSH, where whatever a command prints becomes a graphed, thresholded reading, and over SNMP where the agent is enabled. A Linux host is one device with unlimited sensors, in the same console as the Windows estate.


Does it monitor Windows Server 2012 R2, 2016, 2019 and 2022, and what are the performance monitoring best practices?

Every supported Windows Server version is reached the same way, over WMI, with the same sensors. The practices that matter: watch disk space as a date rather than a percentage, keep down and degraded separate, subscribe to the event logs that matter rather than all of them, and let one slow server delay only its own readings, which the poller does by design.


Can it monitor FTP servers, IIS, Exchange, VPS hosts and hard disk health?

An FTP server is a TCP port sensor plus the host's service sensor; IIS is an HTTP sensor plus its counters over WMI; Exchange is the mail-flow round trip plus its services and queues; a VPS is a server like any other, reached over WMI or SSH wherever it is hosted. Hard disk health comes from the host's SMART and controller data where WMI exposes it and from IPMI and Redfish on servers with a management port.


Is there MSP server monitoring, and what is the server monitoring pricing?

Remote probes (paid editions) let one installation watch several client sites from inside each, and smart groups keep them apart. Pricing is per monitored device with unlimited sensors on a perpetual license, 5 devices free; the licensing page has the rest.


Is there a graphical network monitor for Windows and a network print monitor in it?

The console draws the graphs, the map and the NOC wall view in any browser, so the graphical part is built in. Network printers are monitored over SNMP for toner, paper and error state, and a printer that stops answering is a device-down incident like any other.


Add one server and open its WMI explorer

Download SysPeep, add a Windows server with the credentials you already have, and browse its counters property by property. Up to 5 devices are free at full depth with unlimited sensors; the 30-day trial unlocks every feature except clustering.

Download Free TrialAgentless and extensible