Pulseguard

Infrastructure Monitoring

Every server, watched in real time.

A lightweight agent tracks CPU, memory, network, and disk on every server you connect, so degradation shows up before it becomes an outage.

Start freeSee all products
Metrics tracked per host
4
Agent report interval
30s
Unique keypair per server
Ed25519
Shared credentials
No

Every server, reporting in real time.

The PulseGuard agent streams CPU, memory, network, and disk from every connected host, no matter how many you run.

CPU39%
Memory66%
Network310.0 MB/s
Disk51.0%
Containers18 running
Servers24 connected
How it works

From one command to a live timeline

The agent stays out of the way: it samples, signs, and sends. Everything else happens on the dashboard.

  1. 1

    Connect a server

    Generate a one-time enrollment token from your dashboard and run it once on the host. The agent creates a unique Ed25519 keypair for that server right there, so there are no shared credentials to leak.

  2. 2

    The agent reports in every 30 seconds

    CPU, memory, disk, and network throughput are sampled locally and sent as a signed heartbeat, authenticated with that server's own key so nothing else can post a reading in its name.

  3. 3

    Status updates the moment it changes

    Miss three heartbeats in a row and a server moves to Delayed; miss ten and it's marked Offline. Both are surfaced on your dashboard immediately, not on the next scheduled check.

  4. 4

    Every reading builds your history

    Each sample lands in that server's timeline, so a slow memory leak or a disk filling up shows up as a trend days before it turns into a page.

What's tracked

Four metrics, on every connected host

No configuration required. The agent starts reporting all four the moment it enrolls, on every server you connect.

CPU utilization

Load sampled over a rolling 1-second window before every report, so a brief spike doesn't read as sustained pressure.

cpuPercent

Memory

Used and total bytes reported alongside the percentage, so a tight-but-stable host reads differently from one trending toward OOM.

memPercentmemUsedBytesmemTotalBytes

Disk

Usage tracked as both a percentage and raw bytes: the numbers you actually need before a volume fills up.

diskPercentdiskUsedBytesdiskTotalBytes

Network throughput

Inbound and outbound traffic measured in bytes per second, so a bandwidth spike shows up the same cycle it starts.

netRxBytesPerSecnetTxBytesPerSec

Infrastructure Monitoring, answered.

Every 30 seconds by default: CPU, memory, disk, and network, sent as one signed request. Miss three reports in a row and the server is marked Delayed; miss ten and it's marked Offline.