Installing 4, 8 or even 16 IP cameras is normally straightforward because a modern Gigabit Ethernet network usually has enough capacity to handle the traffic.

The situation changes considerably when the installation grows to:

• 20 cameras
• 32 cameras
• 48 cameras
• 64 cameras
• 100+ cameras
• Multiple NVRs or recording servers
• Multiple buildings or floors
• Central monitoring rooms

At this scale, network bandwidth becomes part of the CCTV system design, not merely an IT consideration.

A camera may be powered correctly, ping successfully and even open through its web interface, yet still suffer from intermittent video loss because the network path between the camera and NVR/VMS cannot reliably transport all video streams.

Typical symptoms include:

• Camera randomly showing Offline
• Video freezing for several seconds
• "No Resource" or similar errors
• Delayed live video
• Missing recorded footage
• Frames being skipped
• Playback becoming jerky
• Cameras reconnecting automatically
• NVR showing network abnormality
• Live view working but playback having gaps
• Problems becoming worse when many cameras detect movement simultaneously

Correct bandwidth planning greatly reduces these problems.

1. First Understand: IP CCTV Is a Data Network

An IP camera is essentially a network device continuously generating video data.

The basic path is:

Camera → PoE Switch → Network/Uplink → Core Switch → NVR/VMS → Storage

For remote viewing another path exists:

NVR/VMS → Router/Firewall → Internet → Remote Client

Therefore, several different bandwidth limits need to be considered.

Do not simply ask:

"Is my switch Gigabit?"

Instead ask:

"How much traffic is being generated, where is it travelling, and what is the maximum capacity of every link and device along that path?"

This distinction is critical.

2. Bitrate Is More Important Than Camera Megapixels Alone

A common mistake is calculating CCTV bandwidth only from camera resolution.

For example:

64 × 4MP cameras

does not tell us the actual bandwidth requirement.

We need to know the bitrate of each stream.

Bitrate may be represented as:

• Kbps
• Mbps
• Mbit/s

For example, one camera may produce:

Main Stream = 4 Mbps

while another 4MP camera might produce:

Main Stream = 8 Mbps

depending on configuration and scene complexity.

Bandwidth is affected by:

• Resolution
• FPS
• Codec
• Compression level
• Scene complexity
• Motion
• Lighting
• Noise
• WDR
• Bitrate mode
• I-frame/GOP configuration
• Smart codec features
• Number of streams requested

Axis specifically notes that VBR traffic changes with scene complexity and activity, while additional streams with different configurations also create additional processing and traffic considerations.

3. Basic CCTV Bandwidth Formula

A useful first calculation is:

Total Camera Bandwidth = Number of Cameras × Average Bitrate per Camera

Suppose:

32 cameras × 4 Mbps = 128 Mbps

For 64 cameras:

64 × 4 Mbps = 256 Mbps

But this should not mean that a 256 Mbps network is sufficient.

You need headroom for:

• Bitrate spikes
• Sub-streams
• Camera management traffic
• Retransmissions
• Network overhead
• Live viewing
• VMS communication
• Future cameras
• Failover conditions

Therefore, design around peak expected traffic plus a safety margin, rather than average bitrate alone.

4. Main Stream + Sub Stream Must Be Considered

Many installations calculate only the recording stream.

That can underestimate actual traffic.

A camera could have:

Main Stream: 4 Mbps
Sub Stream: 512 Kbps

Potential total:

4.5 Mbps approximately

Depending on the NVR/VMS and viewing architecture, multiple streams may be requested.

Hikvision provides a useful example of this principle. For an NVR with 320 Mbps incoming capacity, it states that total incoming traffic includes main stream plus sub-stream. At 4 Mbps main + 1 Mbps sub-stream, 64 channels can fit within 320 Mbps; at 6 Mbps + 2 Mbps, the bandwidth mathematically supports only about 40 such channels.

This demonstrates an extremely important point:

A "64-channel NVR" does not automatically mean that every possible 64-camera configuration can be recorded at any bitrate.

Channel count and bandwidth capacity are separate specifications.

5. NVR Incoming Bandwidth Is Critical

When purchasing an NVR, check:

Incoming Bandwidth / Access Bandwidth

Suppose a 64-channel NVR supports:

320 Mbps incoming bandwidth

If your cameras require:

64 × 4 Mbps = 256 Mbps

there is reasonable capacity available.

But suppose cameras are configured for:

64 × 8 Mbps = 512 Mbps

A 320 Mbps NVR cannot properly accept 512 Mbps of continuous incoming video simply because it has 64 channel licenses.

This can result in:

• Camera disconnection
• Recording interruption
• Stream rejection
• Reduced performance
• Live-view problems

Always check the exact NVR datasheet.

6. Incoming and Outgoing NVR Bandwidth Are Different

NVR specifications may contain values such as:

• Incoming bandwidth
• Outgoing bandwidth
• Total bandwidth

They are not necessarily interchangeable.

Incoming bandwidth

Traffic travelling:

Cameras → NVR

This primarily affects recording capacity.

Outgoing bandwidth

Traffic travelling:

NVR → Viewing Clients

This affects:

• Local monitoring PCs
• Remote users
• Mobile applications
• CMS/VMS
• Multiple simultaneous playback sessions

You can therefore have an NVR that records correctly but becomes slow when many operators simultaneously open cameras.

7. Example Bandwidth Planning Table

The following is an illustrative planning example, not a universal camera specification.

Assume approximately 4 Mbps per camera for the primary stream.

Cameras	Camera Video Traffic	With ~25% Planning Headroom
20	80 Mbps	100 Mbps
32	128 Mbps	160 Mbps
48	192 Mbps	240 Mbps
64	256 Mbps	320 Mbps
80	320 Mbps	400 Mbps
100	400 Mbps	500 Mbps
128	512 Mbps	640 Mbps

This table immediately shows why Fast Ethernet infrastructure should generally be avoided for large CCTV backbones.

8. Why a 100 Mbps Switch Can Become a Problem

Imagine 20 cameras connected across a switch infrastructure.

If:

20 × 4 Mbps = 80 Mbps

it might appear that a 100 Mbps uplink is enough.

In practice, this is poor design.

The link is already operating close to its theoretical maximum.

A sudden increase caused by:

• Heavy movement
• Rain
• Trees moving
• Headlights
• Low-light image noise
• Crowded scenes

can push VBR streams higher.

The result can be congestion and packet loss.

Therefore:

100 Mbps camera port: Often acceptable for an individual camera.

100 Mbps uplink: Usually undesirable for a large camera aggregation switch.

For medium and large CCTV installations, use Gigabit uplinks as a baseline, and consider faster uplinks where aggregation traffic demands it.

9. Do Not Confuse Port Speed With Switch Capacity

A switch having:

24 × Gigabit ports

does not automatically mean it can forward 24 Gbps of real traffic in every direction simultaneously.

Check specifications such as:

• Switching capacity
• Forwarding rate
• Backplane capacity
• Uplink capacity
• Packet forwarding performance

Cheap switches may become bottlenecks in demanding surveillance deployments.

10. Recommended Architecture for 20 Cameras

Example:

Camera 01 ─┐
Camera 02  │
Camera 03  │
...        ├── 24-Port PoE+ Gigabit Switch
Camera 20 ─┘
                  │
             1 Gbps Uplink
                  │
                 NVR

If cameras average 4 Mbps:

20 × 4 = 80 Mbps

A properly engineered Gigabit network provides substantial room above this traffic level.

11. Recommended Architecture for 32 Cameras

Instead of forcing everything onto one inexpensive large PoE switch, consider distributed switching.

16 Cameras
     │
PoE Switch A
     │
     ├──── Gigabit ────┐
                       │
                    Core Switch ─── NVR
                       │
     ├──── Gigabit ────┘
     │
PoE Switch B
     │
16 Cameras

At 4 Mbps:

32 × 4 Mbps = 128 Mbps

Each 16-camera group generates approximately:

64 Mbps

This provides a clean, scalable architecture.

12. Recommended Architecture for 48 Cameras

Example:

16 Cameras → PoE Switch A ─┐
                           │
16 Cameras → PoE Switch B ─┼→ Core Gigabit/10G Switch → NVR
                           │
16 Cameras → PoE Switch C ─┘

At 4 Mbps:

48 × 4 = 192 Mbps

The core link should be designed with ample spare capacity.

13. Recommended Architecture for 64 Cameras

A professional 64-camera deployment could look like:

16 Cameras → PoE Switch A ─┐
                           │
16 Cameras → PoE Switch B ─┤
                           ├→ Core Managed Switch → NVR/VMS
16 Cameras → PoE Switch C ─┤
                           │
16 Cameras → PoE Switch D ─┘

At 4 Mbps:

64 × 4 = 256 Mbps

At 6 Mbps:

64 × 6 = 384 Mbps

At 8 Mbps:

64 × 8 = 512 Mbps

The difference is enormous.

This is why camera count alone should never determine the network design.

14. Consider 10 Gigabit for Larger Aggregation Networks

Gigabit Ethernet is sufficient for many 20–64 camera systems when properly designed.

However, 10GbE becomes useful when:

• Hundreds of cameras exist
• Multiple NVRs are connected
• Multiple recording servers exist
• High-resolution cameras are common
• 4K/8MP cameras are heavily deployed
• Analytics servers consume streams
• Storage traffic shares the same backbone
• Multiple buildings converge into one core
• Failover servers exist
• Multiple monitoring stations request many streams

For example:

Building A Cameras
       │
   PoE Switch
       │
     Fiber
       │
       ├─────────────┐
                     │
Building B Cameras   │
       │             │
   PoE Switch ───────┼→ 10GbE Core
                     │       │
Building C Cameras   │       ├── NVR 1
       │             │       ├── NVR 2
   PoE Switch ───────┘       ├── VMS Server
                             └── Monitoring

15. Use Fiber for Long CCTV Backbone Links

Ethernet over copper has distance limitations.

For long-distance links between:

• Buildings
• Floors
• Warehouses
• Factories
• Campus locations
• Parking areas
• Security rooms

fiber is often preferable.

Advantages include:

• Long distance
• High bandwidth
• Electrical isolation
• Resistance to electromagnetic interference
• Gigabit and 10 Gigabit capability
• Better backbone scalability

Use appropriate:

• SFP
• SFP+
• Single-mode fiber
• Multimode fiber

according to distance and network requirements.

16. H.265 Can Significantly Reduce Bandwidth Requirements

Where the complete system supports it, H.265 can reduce network and storage requirements compared with H.264.

The chain must support the codec:

Camera → NVR/VMS → Client → Playback/Export Environment

Hanwha specifically recommends H.265 as one of the first methods for reducing camera bandwidth on compatible systems.

Do not blindly enable a newer codec without checking compatibility with:

• NVR
• VMS
• Browser/client
• Mobile app
• Analytics
• Third-party integrations

17. Smart Codec Technologies Can Help Further

Different manufacturers provide proprietary bandwidth optimization technologies.

These systems attempt to allocate more data to important parts of an image while compressing less-important portions more aggressively.

For example, Hanwha's WiseStream can reduce bitrate and can work alongside features such as Dynamic GOP/GoV and Dynamic FPS.

Axis similarly provides Zipstream and dynamic FPS/GOP options to reduce bandwidth.

Always test forensic quality after enabling aggressive compression.

18. CBR vs VBR – Which Is Better?

CBR – Constant Bit Rate

The encoder attempts to remain around a defined bitrate.

Advantages:

• Easier network planning
• More predictable storage
• Easier bandwidth control

Potential disadvantage:

Image quality may have to change when scene complexity increases.

VBR – Variable Bit Rate

The bitrate changes according to scene complexity.

Advantages:

• Better ability to preserve image quality
• Efficient allocation of bits

Disadvantage:

• Bandwidth can suddenly increase
• Storage consumption becomes less predictable

Axis notes that VBR can increase significantly with scene complexity and recommends ensuring infrastructure has sufficient margins.

For surveillance networks with constrained links, maximum bitrate controls can be valuable, but excessively restricting bitrate may reduce important image detail.

19. Why All Cameras May Suddenly Increase Bandwidth

Imagine a 64-camera outdoor site at night.

Normally cameras might average:

3 Mbps

Total:

64 × 3 = 192 Mbps

Then heavy rain begins.

The scene becomes much more difficult to compress because of:

• Rain
• Reflections
• Image noise
• Moving trees
• Vehicle headlights
• Shadows
• Constant movement

Individual VBR streams may increase significantly.

If the network was designed around exactly 192 Mbps with little margin, congestion can occur.

This explains why a CCTV system can work perfectly most of the day but fail during:

• Rain
• Night
• Crowds
• Shift changes
• Traffic peaks
• Festivals/events
• High-motion scenes

20. FPS Has a Major Effect on Bandwidth

Many installations unnecessarily configure every camera at:

25 FPS or 30 FPS

Not every surveillance location requires this.

Depending on the security objective:

10–15 FPS may be adequate for many general surveillance applications.

Higher frame rates may be appropriate for:

• Cash counters
• Fast production machinery
• High-speed traffic
• Casinos
• Critical entrances
• Situations requiring detailed motion analysis

Reducing FPS can lower bandwidth and storage, but do not reduce it so far that important events cannot be reconstructed.

21. Resolution Should Match the Actual Security Requirement

Do not configure every camera at maximum resolution simply because the camera supports it.

Examples:

A general corridor may not require the same image detail as:

• Number plate capture
• Cash counter
• Main entrance
• Jewellery counter
• Production inspection point

Higher resolution normally means more pixels to encode and potentially more bandwidth/storage.

Design cameras according to their actual detection, observation, recognition or identification requirement.

22. Configure Main Stream and Sub Stream Properly

A highly effective design is:

Main Stream

Use for recording.

Example:

4MP / H.265 / 15 FPS / 3–5 Mbps

Sub Stream

Use for multi-camera live view and remote monitoring.

Example:

640×360 or similar / H.265 or H.264 / 8–12 FPS / 256–512 Kbps

Instead of requesting 16 full-resolution streams for a 16-camera matrix, the VMS/NVR can display lower-bandwidth sub-streams.

When the operator opens one camera full-screen, the client can switch to the main stream.

This substantially reduces client-side and NVR outgoing traffic.

23. Multiple Monitoring Stations Can Create Unexpected Load

Suppose 64 cameras are recording normally.

Now three security operators each open 64-camera live views.

If the architecture requests additional main streams, bandwidth and processing load can rise sharply.

This is another reason to configure sub-streams correctly.

The system should ideally behave like:

Recording → Main Stream

Multi-camera Live View → Sub Stream

Full Screen → Main Stream

Mobile Remote View → Sub Stream

24. Separate CCTV Traffic From the Office Network

For larger installations, avoid mixing unrestricted CCTV traffic with:

• Employee computers
• Wi-Fi
• Internet browsing
• VoIP
• ERP
• NAS transfers
• Server backups

Prefer a dedicated CCTV VLAN or physical network where appropriate.

Example:

VLAN 10 → Office Computers
VLAN 20 → Servers
VLAN 30 → CCTV Cameras
VLAN 40 → Access Control
VLAN 50 → VoIP

This provides:

• Better traffic isolation
• Easier troubleshooting
• Better security
• Easier monitoring
• Reduced broadcast interaction
• Better policy control

25. Use Managed Switches for Large Installations

For 32, 48, 64+ camera installations, managed switches are highly recommended.

Useful features include:

• VLAN
• QoS
• SNMP
• Port statistics
• Error counters
• Link monitoring
• STP/RSTP
• Link aggregation
• SFP/SFP+ management
• Loop detection
• Port mirroring
• PoE monitoring
• Remote port restart

If a camera repeatedly disconnects, you can examine the switch port for:

• CRC errors
• Packet errors
• Link flapping
• Speed/duplex problems
• Excessive traffic

This can dramatically reduce troubleshooting time.

26. Check PoE Budget – Not Just Bandwidth

A camera can fail even when network bandwidth is perfect.

Every PoE switch has a PoE power budget.

For example:

24 ports may exist, but the total switch power budget could be:

250 W

If cameras collectively demand more than the available power, some ports may become unstable.

Always calculate:

Total Camera Maximum Wattage < Available PoE Budget

Include margin.

Pay special attention to:

• PTZ cameras
• IR cameras
• Heaters
• Illuminators
• PoE+
• PoE++
• Multi-sensor cameras

A camera drawing more power at night when IR LEDs activate can expose a marginal PoE design.

27. Switch Uplink Oversubscription

Consider:

24 cameras × 8 Mbps = 192 Mbps

A Gigabit uplink handles this comfortably under normal conditions.

But imagine multiple switches converging through one restricted link:

Switch A ─┐
Switch B ─┤
Switch C ─┼→ Bottleneck Link → NVR
Switch D ─┘

Even if each access switch works perfectly, the common aggregation link can become overloaded.

Always calculate bandwidth at every aggregation point.

28. Bandwidth Must Be Calculated Link by Link

This is one of the most important design principles.

Do not calculate only:

Total Cameras × Bitrate

Calculate traffic through:

1. Camera Ethernet port
2. Access switch
3. Access-switch uplink
4. Building uplink
5. Fiber link
6. Core switch
7. NVR interface
8. Recording server interface
9. Storage network
10. Monitoring client connection
11. Router/firewall
12. Internet connection

Every path must have enough capacity.

29. Example – 64 Camera Professional Design

Suppose:

64 × 4MP cameras

Configuration:

• H.265
• 15 FPS
• Main stream = approximately 4 Mbps
• Sub-stream = approximately 512 Kbps

Main recording bandwidth:

64 × 4 = 256 Mbps

Potential sub-stream traffic:

64 × 0.5 = 32 Mbps

Potential combined stream load:

288 Mbps

Add 25% design margin:

288 × 1.25 = 360 Mbps

A sensible design could therefore include:

• Gigabit camera aggregation uplinks
• Managed core switch
• NVR rated comfortably above the required incoming traffic
• Gigabit or faster NVR interface
• Separate CCTV VLAN
• Proper sub-stream configuration
• Monitoring of switch and NVR utilization

If substantial expansion is expected, designing the core around 10GbE can provide much greater scalability.

30. Example – 64 × 8MP Cameras

Now suppose:

64 × 8MP cameras

and each averages:

8 Mbps

Recording traffic:

64 × 8 = 512 Mbps

Add 25% margin:

640 Mbps

Now a 1Gbps interface is still mathematically above the calculated stream traffic, but the design is getting much closer to the practical limits once other traffic, spikes and expansion are considered.

This is where you should carefully evaluate:

• 10GbE core
• Multiple NVR interfaces
• Multiple recording servers
• Load distribution
• Camera grouping
• Storage throughput
• Failover architecture

31. Storage Throughput Also Matters

Network bandwidth is only half the problem.

The NVR must write that video continuously to disk.

For example:

512 Mbps ÷ 8 = 64 MB/s

Therefore approximately:

64 MB/s of video data

must be processed continuously before considering additional workloads and overhead.

Playback adds simultaneous read operations.

A system may therefore experience problems because of:

• HDD bottlenecks
• RAID limitations
• NVR CPU
• NVR memory
• Database operations
• Simultaneous playback

rather than Ethernet itself.

32. Bandwidth and Storage Are Directly Related

A useful approximation is:

Storage per day = Mbps × 10.8 GB

Therefore:

4 Mbps ≈ 43.2 GB/day

per camera for continuous recording.

For 64 cameras:

64 × 43.2 ≈ 2.76 TB/day

Actual consumption varies with codec, VBR behaviour, scene complexity and recording schedule.

This illustrates why bitrate optimization affects both:

Network bandwidth + Storage capacity

33. Recommended Utilization Targets

Avoid designing critical CCTV networks to operate continuously at the theoretical maximum.

As a practical engineering approach, try to keep normal sustained traffic comfortably below link capacity and retain headroom for spikes and expansion.

For example:

1Gbps link

Do not intentionally design a critical surveillance backbone around nearly 1Gbps of continuous video.

Once expected sustained traffic moves into the high hundreds of Mbps, evaluate:

• 10GbE
• Link aggregation where correctly supported
• Multiple network interfaces
• Distributed recording
• Multiple NVRs

The exact threshold depends on the architecture and equipment.

34. Do Not Use QoS as a Substitute for Capacity

QoS can prioritize CCTV traffic when CCTV shares infrastructure with other services.

However:

QoS does not create bandwidth.

If a 100 Mbps bottleneck is receiving 180 Mbps of traffic, QoS cannot turn it into a 180 Mbps connection.

Correct solution:

Increase capacity or reduce traffic.

QoS should complement good network design, not compensate for undersized links.

35. Packet Loss Is Dangerous for CCTV

Hanwha's guidance for large surveillance systems specifically emphasizes avoiding packet loss and jitter between cameras and servers because lost streaming data can significantly affect recorded footage.

Packet loss may result from:

• Network congestion
• Bad cable
• Damaged RJ45 connector
• Poor fiber termination
• Defective SFP
• Network loop
• Switch overload
• Duplex mismatch
• EMI
• Bad PoE switch
• Excessive uplink utilization

A camera may remain pingable while the video itself is unreliable.

36. Ping Alone Does Not Prove CCTV Network Health

A common troubleshooting mistake is:

"Camera is pinging, so network is fine."

Incorrect.

Ping uses tiny ICMP packets.

A camera may successfully respond to ping while continuously attempting to send several Mbps of video.

You need to inspect:

• Packet loss
• Latency
• Jitter
• Interface utilization
• CRC errors
• Dropped packets
• Camera bitrate
• NVR incoming traffic
• Switch uplink utilization

37. Avoid Network Loops

Large CCTV systems often have multiple switches.

Incorrect cabling can create a loop:

Switch A ───── Switch B
   │              │
   └──────────────┘

Without proper STP/RSTP configuration, this can create broadcast storms and effectively bring down the CCTV network.

Managed switches and correct spanning-tree design are strongly recommended.

38. IP Address Planning Is Important

Avoid randomly assigning IP addresses.

For example:

192.168.30.1       Gateway
192.168.30.10      NVR
192.168.30.11      Core Switch
192.168.30.20-29   Access Switches
192.168.30.50-149  Cameras

Keep documentation containing:

• Camera number
• Camera name
• Location
• IP address
• MAC address
• Switch
• Port number
• Resolution
• Main bitrate
• Sub bitrate
• FPS
• Codec
• Firmware version

This becomes extremely valuable during troubleshooting.

39. Avoid Duplicate IP Addresses

Duplicate IP addresses can cause strange intermittent CCTV failures.

Symptoms include:

• Camera available sometimes
• Camera suddenly offline
• Wrong camera appearing
• NVR authentication errors
• Ping responses changing
• Stream randomly stopping

Use:

• Static addressing
• DHCP reservations
• IP address management

but avoid uncontrolled mixtures of static and dynamic addresses.

40. Monitor Actual Camera Bitrates

Do not rely only on configured bitrate.

Monitor actual traffic.

Modern camera/VMS platforms may expose:

• Current bitrate
• Average bitrate
• Peak bitrate
• Historical bitrate

Hanwha, for example, provides reporting for average streaming bitrate and mechanisms to set maximum streaming bitrate in supported systems.

Also monitor switch ports using:

• SNMP
• Network monitoring software
• Switch web interface
• VMS statistics

41. Night-Time Testing Is Essential

A CCTV installation should not be declared stable only because it works at 2 PM.

Test:

• Day
• Night
• IR enabled
• Rain if possible
• High traffic
• Many moving people
• All cameras live
• Simultaneous playback
• Remote viewing

Night-time noise and IR activity can significantly change compression requirements.

42. Recommended Camera Configuration Starting Point

For a typical general-surveillance 4MP camera, a reasonable starting configuration might be:

Resolution: 4MP
Codec: H.265
FPS: 15
Bitrate: approximately 3–5 Mbps
Smart Codec: Enabled where appropriate
Main Stream: Recording
Sub Stream: Live View
Sub Stream FPS: 8–12
Sub Stream Bitrate: approximately 256–512 Kbps

These are starting points only.

High-security applications may require higher settings.

Never reduce bandwidth so aggressively that:

• Faces become unclear
• Number plates disappear
• Motion becomes unusable
• Important forensic evidence is lost

43. Suggested Network Approach by Camera Count

Camera Count	Recommended Network Approach
1–8	Gigabit switch preferred
9–16	Gigabit PoE switch
17–24	Gigabit managed PoE switch
25–32	Multiple managed PoE switches + Gigabit core
33–48	Distributed PoE + managed core + Gigabit/fiber uplinks
49–64	Managed aggregation architecture + Gigabit/fiber; evaluate 10GbE core
65–128	10GbE core increasingly valuable
128+	Enterprise VMS/network architecture and distributed recording should be evaluated

These are architectural guidelines rather than rigid limits.

Actual design must be based on calculated traffic.

44. Bandwidth Worksheet Before Installation

Before purchasing hardware, create a table like this:

Camera	Resolution	FPS	Codec	Main Mbps	Sub Mbps	Switch
CAM01	4MP	15	H.265	4	0.5	SW1
CAM02	4MP	15	H.265	4	0.5	SW1
CAM03	8MP	15	H.265	8	0.5	SW1

Then calculate:

Per-switch traffic

Per-uplink traffic

Core traffic

NVR incoming traffic

Monitoring traffic

This prevents guesswork.

45. A Better Formula for Professional CCTV Planning

Instead of:

Cameras × Bitrate

use:

Required Bandwidth = Σ(Main Streams + Required Additional Streams) × Safety Factor

For example:

64 cameras:

Main:

4 Mbps

Sub:

0.5 Mbps

Combined:

4.5 Mbps

Total:

64 × 4.5 = 288 Mbps

With 25% headroom:

288 × 1.25 = 360 Mbps

Then verify that every relevant network path and NVR/VMS interface can support this load.

46. What Happens When Bandwidth Is Insufficient?

When network capacity becomes exhausted, symptoms may include:

Stage 1

• Increased latency
• Delayed live view

Stage 2

• Packet queues
• Jitter

Stage 3

• Packet drops
• Video artifacts

Stage 4

• Stream timeout
• Camera disconnected by NVR

Stage 5

• Recording gaps
• Multiple cameras appear offline

This is why bandwidth shortage is sometimes incorrectly diagnosed as defective cameras.

47. Camera Offline Does Not Always Mean Camera Failure

If 5–10 cameras simultaneously disconnect, ask:

Are they connected to the same switch?

Do they share the same uplink?

Are they on the same PoE power supply?

Do they travel through the same fiber/SFP?

If several cameras fail simultaneously, the common network component is often more suspicious than all cameras independently failing.

48. Troubleshooting Camera Signal Loss

When cameras randomly disappear:

Step 1 – Check Power

Inspect PoE consumption and budget.

Step 2 – Ping Camera

Check latency and packet loss.

Step 3 – Check Camera Bitrate

Compare configured and actual values.

Step 4 – Check Switch Port

Look for:

• CRC errors
• Drops
• Link flapping

Step 5 – Check Switch Uplink

Look for excessive utilization.

Step 6 – Check NVR Incoming Bandwidth

Compare actual traffic with NVR specification.

Step 7 – Check NVR CPU/RAM

High utilization can cause symptoms similar to network failure.

Step 8 – Check Storage

Slow or failing disks can cause recording problems.

Step 9 – Check Cabling

Use certified Cat5e/Cat6 cabling and test questionable runs.

Step 10 – Test During Peak Activity

Do not troubleshoot only during quiet periods.

49. Large CCTV System Reliability Principles

For 32–64+ cameras, a professional design should consider:

Capacity

Enough bandwidth.

Headroom

Enough unused bandwidth for peaks.

Segmentation

Dedicated VLAN/network.

Redundancy

Avoid unnecessary single points of failure.

Monitoring

Watch traffic and errors.

Documentation

Maintain camera/switch/IP mapping.

Scalability

Allow future expansion.

Power

Adequate PoE and UPS capacity.

Storage

Enough write throughput and retention capacity.

Security

Restrict camera network access.

Hanwha's large-system guidance similarly emphasizes network throughput, loss/jitter avoidance, continuous monitoring and failure-recovery planning.

50. Practical Design Example – 64 Cameras

A strong architecture could be:

                    CCTV VLAN
                       │
                Managed Core Switch
                 /      |      \
                /       |       \
         Gigabit     Gigabit    Gigabit/Fiber
            │           │            │
        PoE SW-1     PoE SW-2      PoE SW-3
        16 CAM       16 CAM        16 CAM
                                     
                       │
                    PoE SW-4
                    16 CAM
                       │
                       │
                Managed Core
                       │
                  NVR / VMS
                       │
                    Storage

For a more demanding environment:

Access PoE Switches
       ↓
Gigabit/Fiber Uplinks
       ↓
10GbE Core
       ↓
NVR/VMS Recording Servers
       ↓
High-Performance Storage

This provides a much better foundation for expansion.

51. Important Rule: Design for Bandwidth, Not Just Camera Count

Two 64-camera installations can have completely different requirements.

Installation A

64 × 2MP

15 FPS

H.265

2 Mbps

Total approximately:

128 Mbps

Installation B

64 × 8MP

25 FPS

Higher quality

8 Mbps

Total approximately:

512 Mbps

Both have:

64 cameras

But Installation B requires approximately four times the video bandwidth in this simplified example.

Therefore:

Camera count is not a bandwidth specification.

52. How Proper Bandwidth Planning Reduces Camera Failure

Correct bandwidth engineering helps reduce:

• Random camera offline errors
• Video freezing
• Missing recordings
• Packet loss
• High latency
• Playback gaps
• NVR overload
• Switch congestion
• Unstable remote viewing
• Delayed camera streams

It does not prevent actual hardware failures, but it removes one of the most common infrastructure causes of apparent "camera signal failure."

53. Recommended Design Checklist

Before finalizing a 20, 32, 48 or 64+ camera installation, verify:

1. Camera resolution
2. Camera FPS
3. Codec
4. Main-stream bitrate
5. Sub-stream bitrate
6. Smart codec
7. Camera PoE requirement
8. Switch PoE budget
9. Switch port speed
10. Switch backplane/switching capacity
11. Uplink speed
12. Fiber requirements
13. Core-switch capacity
14. NVR incoming bandwidth
15. NVR outgoing bandwidth
16. NVR channel capacity
17. NVR NIC speed
18. Storage write performance
19. Recording retention requirement
20. VLAN design
21. IP addressing
22. Monitoring requirements
23. UPS backup
24. Expansion capacity
25. Failover requirements

54. Final Recommended Approach

For a modern medium-to-large CCTV installation:

Camera Layer

Use properly configured H.265/H.264 cameras with sensible resolution, FPS and bitrate.

↓

Access Layer

Use quality Gigabit managed PoE switches with sufficient PoE budget.

↓

Aggregation Layer

Use Gigabit or fiber uplinks with substantial headroom.

↓

Core Layer

Use a managed Gigabit core for moderate installations and consider 10GbE as aggregate traffic and future growth increase.

↓

Recording Layer

Select NVR/VMS capacity according to:

Channel Count + Incoming Bandwidth + Outgoing Bandwidth + Storage Throughput

↓

Monitoring

Continuously monitor:

Bitrate + Packet Loss + Port Errors + Uplink Utilization + PoE Load + NVR Performance

Conclusion

When designing a CCTV installation with 20, 32, 48, 64 or more IP cameras, bandwidth should be calculated before selecting the NVR and network infrastructure.

The fundamental calculation is:

Camera Count × Camera Bitrate = Base Video Bandwidth

But professional design goes further:

Total Required Bandwidth = Main Streams + Additional Streams + Network Overhead + Peak-Traffic Headroom

The most important rule is:

Never design a CCTV network to operate continuously near its maximum capacity.

A good surveillance network should have sufficient spare capacity to absorb bitrate variations, simultaneous live viewing, network overhead, future cameras and unexpected high-motion scenes.

Correctly combining:

H.265/appropriate codec + sensible bitrate + suitable FPS + main/sub-stream design + managed PoE switches + Gigabit/fiber uplinks + adequate NVR bandwidth + sufficient storage performance + network monitoring

can dramatically improve the reliability of a large IP CCTV system.

FAQ

1. How much bandwidth does one CCTV camera require?

There is no universal value. It depends on resolution, FPS, codec, compression, scene complexity and camera configuration. A camera might use 2 Mbps, 4 Mbps, 8 Mbps or substantially more depending on requirements.

2. How much bandwidth do 32 cameras require?

If each uses 4 Mbps:

32 × 4 = 128 Mbps

Additional traffic and safety margin must then be considered.

3. How much bandwidth do 64 cameras require?

At 4 Mbps each:

64 × 4 = 256 Mbps

At 8 Mbps each:

64 × 8 = 512 Mbps

This demonstrates why the bitrate must be known.

4. Is a Gigabit switch enough for 64 cameras?

It can be, depending on bitrate, topology and traffic paths. Do not decide based on camera count alone.

5. Should I use a 10GbE switch for 64 cameras?

Not necessarily, but it can be valuable when using high-resolution/high-bitrate cameras, multiple NVRs, multiple VMS servers, large-scale analytics or when significant expansion is planned.

6. Can low bandwidth make cameras appear offline?

Yes. Congestion, packet loss and stream timeouts can cause cameras to appear offline even when the cameras themselves are functional.

7. Why do cameras work during the day but fail at night?

IR activation, low-light noise and complex scenes can increase encoded video requirements. PoE consumption may also increase when IR illuminators activate.

8. Is H.265 better for large CCTV installations?

It can substantially reduce bandwidth/storage compared with H.264 in compatible environments. Confirm that cameras, NVR/VMS, clients and integrations support it.

9. What is NVR incoming bandwidth?

It is the maximum video/data rate the NVR can accept from connected cameras.

10. What is NVR outgoing bandwidth?

It is the capacity available for sending streams from the NVR to monitoring clients, remote users or other systems.

11. Can a 64-channel NVR always record 64 cameras?

Not necessarily at every possible bitrate. The NVR must support both the required number of channels and their combined incoming bandwidth. Hikvision explicitly distinguishes channel count from access/incoming bandwidth.

12. Should CCTV use CBR or VBR?

Both have uses. CBR/limited bitrate provides predictability, while VBR can preserve quality by allowing bitrate to change with scene complexity. Network design must account for possible VBR peaks.

13. What is CCTV sub-stream?

A sub-stream is a lower-resolution/lower-bitrate stream commonly used for multi-camera live viewing and remote/mobile monitoring.

14. Should recording use the main stream?

Normally yes. The main stream generally provides the higher-quality recording required for evidence.

15. How much network capacity should remain unused?

There is no universal percentage, but maintaining meaningful headroom is important. A practical starting design margin such as 20–30% can be useful, with larger margins for critical or highly variable systems.

16. Can a PoE switch cause camera disconnection?

Yes. Insufficient PoE budget, overheating, bad ports, overloaded uplinks and hardware problems can cause intermittent cameras.

17. Why do multiple cameras disconnect simultaneously?

Look for common infrastructure: the same PoE switch, uplink, SFP, fiber path, power supply, VLAN or NVR interface.

18. Is ping enough to test CCTV network quality?

No. Ping does not measure sustained video-stream performance. Check packet loss, bitrate, port errors, link utilization and NVR/VMS statistics.

19. Should CCTV be placed on a separate VLAN?

For medium and large installations, it is generally a strong design practice because it improves isolation, management, troubleshooting and security.

20. Can QoS fix an overloaded CCTV network?

No. QoS prioritizes traffic; it does not create additional bandwidth.

21. Does lowering FPS reduce bandwidth?

Generally yes, but the effect varies with codec and scene. Do not reduce FPS below the application's evidentiary requirements.

22. Does lowering resolution reduce bandwidth?

Generally yes, because fewer pixels need to be encoded, although actual bitrate still depends on encoder settings and scene complexity.

23. What FPS should CCTV cameras use?

There is no universal value. Around 10–15 FPS can be suitable for many general surveillance scenes, while high-motion or specialized applications may require higher frame rates.

24. Can bad Cat6 cable cause video loss?

Yes. Bad termination, excessive cable length, damaged cable, EMI and poor connectors can produce errors and packet loss.

25. Is fiber better for large CCTV installations?

Fiber is highly useful for long-distance and backbone connections because it provides high bandwidth, electrical isolation and resistance to electromagnetic interference.

26. Should I use managed or unmanaged switches?

Managed switches are strongly preferable for larger installations because VLAN, SNMP, port statistics, STP/RSTP, PoE monitoring and diagnostics improve reliability and troubleshooting.

27. Can storage cause CCTV video freezing?

Yes. An overloaded or failing storage subsystem can cause recording and playback issues even when network bandwidth is adequate.

28. What is the biggest mistake in large CCTV installations?

Selecting equipment based only on camera/channel count without calculating bitrate, uplink capacity, NVR bandwidth, PoE budget and storage throughput.

29. Should all cameras use maximum resolution?

No. Configure resolution according to the actual surveillance requirement.

30. What is the best way to design a reliable 64-camera CCTV system?

Calculate every stream, divide cameras across quality managed PoE switches, use appropriately sized Gigabit/fiber uplinks, provide a capable core network, verify NVR incoming/outgoing bandwidth, use sub-streams intelligently and maintain significant capacity headroom.

 

#CCTV #CCTVBandwidth #IPCamera #IPCameraBandwidth #CCTVNetwork #NetworkCCTV #NVR #NVRBandwidth #CCTVInstallation #CCTVSetup #CCTVDesign #CCTVNetworking #SecurityCamera #VideoSurveillance #SurveillanceSystem #PoESwitch #GigabitSwitch #ManagedSwitch #CCTVPoE #H265 #H264 #VideoCompression #CameraBitrate #CCTVBitrate #CCTVStorage #CCTVTroubleshooting #CameraOffline #VideoLoss #PacketLoss #NetworkCongestion #CCTVVLAN #NetworkDesign #CCTVInstaller #CCTVTechnician #CCTVEngineer #CCTVGuide #CCTVTechnicalGuide #CCTVInfrastructure #CCTVFiber #FiberNetwork #10GbE #GigabitEthernet #CCTVMonitoring #SecuritySystem #NetworkSecurity #CCTVOptimization #CCTVFailure #CCTVMaintenance #LargeCCTVSystem #VideoSecurity