Network latency spikes at peak hours – quick fix

When it hits and what triggers it

You're in a meeting at 2:15 PM and the VPN drops. Or your remote desktop freezes. This happens every weekday, same time, like clockwork. The culprit is almost always a scheduled task — a backup job, a cloud sync (OneDrive, Dropbox, Google Drive), or a software update that kicks off at the same time. I've seen this on Windows Server 2019, Office 365 backups, and even a Linux box running rsync at 2 PM sharp.

Why it happens

Your network pipe is big enough for regular traffic — email, web browsing, a few video calls. But when a backup or sync starts, it grabs as much bandwidth as it can. That saturates the link. Latency goes from 10ms to 500ms+. Voice and video codecs can't handle that. They drop packets, and you get lag.

The root cause is simple: no traffic shaping or QoS, and no bandwidth caps on those background jobs. Most apps default to "use whatever's available". That's fine for a home network. In an office with 50 people, it's a disaster.

How to fix it — step by step

Find the spike machine. Open your router's traffic logs or use a free tool like ntopng (Linux) or NetFlow Analyzer (Windows). Look for the IP with the most traffic at 2 PM. Write it down.
Check that machine's scheduled tasks. On Windows, run taskschd.msc and look under Task Scheduler Library. Sort by Triggers. Any task starting between 1:50 PM and 2:10 PM? Click it, see what it runs. On Linux, check cron with crontab -l or ls /etc/cron.d/.
Reschedule it. Move the job to a time when few people are online — like 10 PM. In Task Scheduler, right-click the task, Properties, Triggers tab, edit. Pick a new time. On Linux, change the cron line.
If you can't move it, cap its bandwidth. On a Windows Server, use the built-in QoS Policy in Group Policy (Computer Configuration > Windows Settings > Policy-based QoS). Set a limit like 5 Mbps for that app. On a Linux server, use tc (traffic control) to shape the traffic. Here's a quick example for a backup app on port 443:
```
tc qdisc add dev eth0 root handle 1: htb default 30
tc class add dev eth0 parent 1: classid 1:1 htb rate 5mbit
```
This limits the backup traffic to 5 Mbps.
Turn on QoS on your router. If you have a business-grade router (MikroTik, pfSense, Ubiquiti), enable QoS. Prioritize voice and video traffic (usually ports 80, 443, 3478-3481). That way, even if a backup runs, latency stays low for critical apps.

Still lagging? Check these

Multiple backups at once. Maybe more than one machine starts at the same time. Use staggered start times — 2:00, 2:15, 2:30.
Cloud sync loops. OneDrive or Dropbox can get stuck in a sync loop if a file is constantly changing (like a database). Pause the sync app manually and see if the spike stops.
ISP throttling. Some ISPs shape traffic during peak hours. Run a speed test at 2 PM and compare to 2 AM. If speeds drop more than 30%, call your ISP.
Old switch or cable. A bad cable can cause retransmits, which look like latency. Check the cable between the backup machine and the switch. Replace it if it's Cat5 or older.

Nine times out of ten, it's the backup. Fix that, and the office runs smooth again.