SystemMatchStarter
For one company or a small security integrator.
£99 / month
- 1 monitored site
- Up to 50 monitored devices
- Up to 2 monitoring agents
- Unlimited users
- Alerts, Insight and daily PDF reports
SystemMatch Monitor watches the things that fail silently — the stopped recording service, the array filling up, the degraded RAID set — and tells somebody before the footage anyone needs turns out not to exist.
No card. Nothing to uninstall from your customer's network but one service.
Here is the same recording server, at the same moment, seen two ways. One of these is what your monitoring is currently telling you.
Every one of those failures leaves the machine up, reachable and green. That is why this product exists: not to tell you a device is online, but to tell you the system is doing its job.
Live checks are available now. Limited scope checks also work, but only on the platforms named here. Roadmap items cannot be selected in the console.
| Check | What it catches | Status |
|---|---|---|
| Ping, ports, web | Is it reachable at all. ICMP with packet loss, TCP connect, and HTTP with an expected status and content match. | Live |
| Windows services | The stopped recording service. Knows the difference between stopped, disabled, paused and mid-restart — a service restarting during patching is a warning, not a 3am call. | Live |
| Disk space | The array that fills up and quietly stops keeping footage. Measures what an application can actually write, not the raw total — a volume can look 5% free to df and be completely full to the service writing to it. | Live |
| CPU and memory | Sustained load, never a single spike. A backup or an antivirus scan pins a processor for a minute and neither is a fault. Memory is judged on what is available, not what is "used". | Live |
| RAID health | Reports how many more disks the array can lose. A degraded RAID6 still has margin; a degraded RAID5 has none, and calling both "degraded" hides the difference. Linux software RAID only — see below. | Limited scope |
| Windows event log | Counts entries matching a level, provider and event id — the disk errors nobody reads. Counts and event ids only; message bodies never leave the customer's network. | Live |
| Days until full | A least-squares projection over measurements already collected. Refuses to give a date when the data does not support one, and shows why instead. | Live |
| Alerting by email | One problem is one email, not one a minute. Each carries what it means and what to do about it. Encrypted in transit, and it never contains a credential. | Live |
| SQL Server health | Backup age, log file growth, failed agent jobs. Not built. | Roadmap |
| SNMP, VMS integrations | Switches, UPS, PoE draw; Milestone and Genetec vendor health. Not built. | Roadmap |
Alerts are written for somebody reading a phone at 3am. This is a real one, not a mock-up:
You are being asked to install software on a customer's recording server. These are the limits, stated up front.
Windows services, CPU, memory, disk space and filtered event-log counts have run successfully on a real 64-bit Windows machine as well as through the automated test suite. Windows Server editions and ARM64 builds still need acceptance on representative customer hardware, so those limits remain explicit rather than being assumed.
Reading a Dell PERC or an HP Smart Array needs a vendor command-line tool. Our agent cannot run a command at all — no part of it can execute anything, and that is checked on every build. We are not quietly spending that guarantee to tick a box. Linux software RAID is read properly; hardware controllers report unknown, never a false healthy.
No addresses, hostnames, credentials, camera or cardholder details, network diagrams or logs. Not by default, not as an option — the software contains no AI client and holds no key, and a build gate fails if one ever appears. Every state, alert, score and projection is produced by deterministic code on your own server.
No per-user charge, no setup fee and no feature is hidden behind a higher tier. Prices exclude VAT.
For one company or a small security integrator.
For multi-site estates and growing integrators.
For larger estates, MSPs and customer portfolios.
Launch pricing is billed monthly and may change for new customers. Existing agreed subscriptions keep their contracted price. The 30-day trial needs no card.
The trial is the full product, not a cut-down version. When it ends nothing is deleted and nothing stops being monitored — you simply cannot add more until you choose a plan.
Email hello@systemmatch.co.uk using the prepared request. We create the company and email its first owner a secure, one-time setup link so they can choose their own password. The owner can then invite administrators, operators and viewers from inside the console. No password is sent by email.
In the console: add a site, register an agent, and you are given a one-time enrolment token. It is shown once — we keep only a hash of it.
Anywhere on the customer's network that stays on. It checks the connection before it installs anything, registers a service, and waits for the first result so you know it worked before you leave site.
Paste a CSV of names, addresses and checks. You see every row and every error before anything is created, and it is all-or-nothing on commit.
One service. Outbound HTTPS only — it opens no port and accepts no connection.
sha256sum systemmatch-agent-linux-amd64.tar.gz —
it must match the hash above, which is also published at
SHA256SUMS.