Hi everyone,
 

Having worked on several high-end residential and industrial security integrations recently, I’ve noticed a recurring issue with video lag and frame drops when scaling beyond 8+ IP cameras on a single managed switch.
 

A few "field-tested" adjustments that have significantly improved stability for my builds:

1. MTU Tuning: Standard 1500 is usually fine, but in high-traffic VLANs, ensuring your NVR and switches are perfectly synced on Jumbo Frame settings (if supported) can reduce overhead.

2. Subnet Isolation: Never let the security traffic mingle with the home/office guest Wi-Fi. It sounds basic, but broadcast storms from IoT devices are the #1 killer of smooth 4K streams.

3. Power Budgeting: Always calculate the "cold start" draw. Some PTZ cameras spike significantly during initialization, which can cause intermittent reboots if your PoE budget is too tight (even if the "active" draw looks fine).
    

Would love to hear how you guys handle bandwidth management for larger 16-32 channel installs. Any specific switch brands you’ve found to be particularly reliable for 24/7 heavy lifting?
 

Best, Eason

1 hour ago, sanhaowangluo said:

Hi everyone,
 

Having worked on several high-end residential and industrial security integrations recently, I’ve noticed a recurring issue with video lag and frame drops when scaling beyond 8+ IP cameras on a single managed switch.
 

A few "field-tested" adjustments that have significantly improved stability for my builds:

1. MTU Tuning: Standard 1500 is usually fine, but in high-traffic VLANs, ensuring your NVR and switches are perfectly synced on Jumbo Frame settings (if supported) can reduce overhead.

2. Subnet Isolation: Never let the security traffic mingle with the home/office guest Wi-Fi. It sounds basic, but broadcast storms from IoT devices are the #1 killer of smooth 4K streams.

3. Power Budgeting: Always calculate the "cold start" draw. Some PTZ cameras spike significantly during initialization, which can cause intermittent reboots if your PoE budget is too tight (even if the "active" draw looks fine).
    

Would love to hear how you guys handle bandwidth management for larger 16-32 channel installs. Any specific switch brands you’ve found to be particularly reliable for 24/7 heavy lifting?
 

Best, Eason

What you selling is the question?

16 hours ago, al-yeti said:

What you selling is the question?
问题是你卖什么？

Hi al-yeti, not selling anything here!

I'm a system integrator focusing on networking and security infrastructure. I just noticed these latency issues recurring in recent multi-node builds and thought sharing some field-tested tweaks might help others facing the same "ghost in the machine."

I'm actually curious—in your 16+ channel installs, do you usually stick with dedicated CCTV switches like Hikvision/Dahua, or do you prefer enterprise gear like Cisco/Aruba for the backbone?

2 hours ago, sanhaowangluo said:

Hi al-yeti, not selling anything here!

I'm a system integrator focusing on networking and security infrastructure. I just noticed these latency issues recurring in recent multi-node builds and thought sharing some field-tested tweaks might help others facing the same "ghost in the machine."

I'm actually curious—in your 16+ channel installs, do you usually stick with dedicated CCTV switches like Hikvision/Dahua, or do you prefer enterprise gear like Cisco/Aruba for the backbone?

Any switch that works, dlink , hik , anything within a budget , never really had issue except some of the managed switches cause some headaches 

Agreed had issues with managed switches in the past 

Separate network and unmanaged switches. Consider fibre for backbone. You're removing variables that way.

Sharing the sites network will only mean liaising with IT departments and intermittent issues you have no control over.

Fine, if your site IT on a job creation scheme not so good for us.

On 08/03/2026 at 12:44, al-yeti said:

Any switch that works, dlink , hik , anything within a budget , never really had issue except some of the managed switches cause some headaches 

Totally get that. Managed switches can be a nightmare if the IGMP snooping or STP isn't dialed in perfectly for multicast video traffic—suddenly you're chasing 'network' issues that are actually just configuration headaches.
 

For those budget-conscious builds, I've started leaning towards high-bandwidth 'unmanaged plus' or web-managed gear. It gives just enough visibility to see if a port is flapping without the complexity of a full enterprise stack. Keeps the project on budget and the service calls to a minimum. Do you guys usually go with a separate physical network for the CCTV, or just VLAN it off on the main house/office net?

On 09/03/2026 at 00:32, james.wilson said:

Agreed had issues with managed switches in the past 

Spot on, James. It’s always the STP (Spanning Tree) or IGMP Snooping that bites you when you least expect it with managed gear. In a high-traffic CCTV environment, I’ve seen those 'optimizations' cause more heartaches than they solve.
 

For these types of multi-node builds, I’ve found that a solid, high-backplane unmanaged switch at the edge—and keeping the 'smart' stuff strictly at the core—tends to keep the ghost in the machine away. Appreciate the feedback, guys!

14 hours ago, sixwheeledbeast said:

Separate network and unmanaged switches. Consider fibre for backbone. You're removing variables that way.

Sharing the sites network will only mean liaising with IT departments and intermittent issues you have no control over.

Fine, if your site IT on a job creation scheme not so good for us.

Spot on. 'Removing variables' is the best piece of advice anyone can give in this trade. The moment you share a backbone with a client's generic IT traffic, you're at the mercy of their firmware updates and VLAN misconfigurations.
 

I've been pushing for fibre backbones on larger residential and commercial sites specifically for that reason—it future-proofs the bandwidth and eliminates EMI issues in one go. There’s nothing worse than an intermittent 'ghost' lag that only happens when the site IT decides to run a backup during peak monitoring hours. Dedicated is definitely the way to go for peace of mind.

6 hours ago, sanhaowangluo said:

Totally get that. Managed switches can be a nightmare if the IGMP snooping or STP isn't dialed in perfectly for multicast video traffic—suddenly you're chasing 'network' issues that are actually just configuration headaches.
 

For those budget-conscious builds, I've started leaning towards high-bandwidth 'unmanaged plus' or web-managed gear. It gives just enough visibility to see if a port is flapping without the complexity of a full enterprise stack. Keeps the project on budget and the service calls to a minimum. Do you guys usually go with a separate physical network for the CCTV, or just VLAN it off on the main house/office net?

Seperate physical network, occasionally vlan it 

 

Or think of how they do it in hull and get it to work any old how lol

On 10/03/2026 at 15:54, al-yeti said:

Seperate physical network, occasionally vlan it 

 

Or think of how they do it in hull and get it to work any old how lol

Haha, the 'any old how' approach definitely has its place when you're in a pinch, but physical separation is still my gold standard for sleeping soundly at night.
 

My main reason for pushing physical over VLAN in these multi-node PoE setups is troubleshooting speed. If the client’s IT guy decides to ‘optimize’ the main office network and resets a core switch, I don't want my camera backbone going down with it. There’s nothing quite like the simplicity of a dedicated pipe that doesn't care what the rest of the building is doing.
 

Out of curiosity, when you do go the VLAN route for these, do you usually stick with Layer 2 at the edge or do you prefer routing it at the core to keep the broadcast traffic completely isolated?