IP surveillance is changing. This is due to the evolution from simple sequential image streams to applying AI to video. However, it is also driven by the increasing migration to the use of the cloud.
What is IP Surveillance?
Images identify suspicious movements. However, they can also show faulty machines and unexpected situations. Therefore, IP surveillance is very important for the industry. From there, you can control the video camera from any IP network. This is the case of LAN, Intranet, or Internet.
However, it is also possible to install the IP camera with conventional wires ( network cable ) for audio and video streaming. Wireless technology with encryption is also compatible with the model. In other words, an IP (Internet Protocol) surveillance system is easy to install. And even easier to maintain. Unlike old analog cameras, which require proprietary cables, tools, and technicians.
However, IP surveillance systems require large amounts of storage media. After all, most surveillance is based on capturing images. This also means that input/output (I/O) will generally be sequential. Think of one frame of film after another being written to or read from a drive. However, this changes somewhat with the use of AI (artificial intelligence) to add intelligence to IP surveillance.
But what are the storage needs of surveillance systems? What type of storage media can be used? And what types of storage products are intended for surveillance?
IP surveillance is changing
Monitoring and retaining video camera footage is still core to the world of surveillance. Therefore, a fundamental requirement of storage in surveillance is the ability to handle sequential I/O well.
Meanwhile, surveillance is increasingly being supplemented with AI. Its name could also be vision intelligence. After all, it aims to bring actionable insights to images captured by cameras.
So, before security cameras would simply be monitored by humans and kept as a record. Now, movements of potential interest can be identified and alerts triggered. This is critical for physical security. After all, systems can recognize an intruder by:
- Movement and more.
The role of AI in IP surveillance
However, AI is also used in other surveillance applications. This is the case of:
- Changing traffic lights to improve vehicle flow;
- Monitoring retail center movement to adjust product placement;
- Ability to recognize well-being concerns in healthcare. For example – like a patient who fell.
However, it doesn’t matter the application and the level of AI applied to it. Sequential AE/O will likely be the primary storage performance characteristic needed for surveillance systems. However, not all seemingly sequential I/O is so smooth when it comes to video.
Some video codecs understand different frame types. These represent complete images and differential changes that contain only parts of the changes. A slower image would see little change. Unless a person walks into the camera’s view. That is, the I/O flow would be quite similar to a wave.
Therefore, image capture requirements are largely sequential. However, applying analytics and machine learning functionality will likely involve an entire payload with more randomness.
This is because the image is quite sequential. However, reading and comparing patterns within images will reference potentially large amounts of existing patterns already saved. That is, the IoT security side is potentially quite random in its I/O needs.
What are the best storage media for IP surveillance?
HDD spinning disk
Sequential AE/O forms the vast majority of IP surveillance traffic. Therefore, spinning disk HDDs are prevalent in the area. After all, the mechanical nature of the drives means time is saved when the read heads don’t have to move in and out of different locations on the disk platters.
Drives can be SATA or the higher performance and more expensive series-attached SCSI (SAS). At least in terms of backend connectivity. However, both will work well with the main task of video ingestion.
Solid State and NVMe
However, for the kind of heavy lifting that analysis requires, solid-state storage does not incur a time penalty. After all, random access to storage cells is done electronically, rather than mechanically.
This type of media means Flash or NVMe Flash. But, you may need enhanced processing power from a GPU card. Solid state is much more expensive than a spinning disk. So it generally makes sense to have a flash layer where the data being processed is kept. Meanwhile, HDD is suitable for mass video storage.
Tape is also a potentially usable mass storage medium. Furthermore, when not in use, they do not consume energy, unlike HDDs. However, data access is much slower than other media. Therefore, automating this through the use of LTFS would be an option. This is a way to layer a NAS-type file system on top of tape. Using tape as another layer for mass storage is another way to incorporate it.
Finally, there’s cloud storage. In some senses, it could be an ideal medium for video surveillance filming and analysis. After all, it can be leveraged in the form of cloud-as-a-service offerings.
However, there are also disadvantages. This is the case with the bandwidth required to upload video. After all, this can multiply as the number of cameras increases. There are also charges for retaining data in the cloud. Compliance can also be an issue. Especially if images are stored outside of specific jurisdictions.
On-site IP surveillance and storage
The easiest way to get IP surveillance is to purchase a network video recorder (NVR). After all, it comes with everything you need:
- Servers ;
- Connection hardware.
Smaller companies can take advantage of it. However, there are limitations regarding:
- Integration possibilities.
When it comes to HDDs for surveillance storage, any corporate drive will work. However, retention of large quantities of images for long periods of time should be considered. Especially with potentially small amounts of access and higher capacities with less stringent performance requirements. After all, that would probably represent 7,200 rpm SATA most of the time.
Increasingly, IP surveillance is moving to the cloud. In so-called video surveillance as a service, the customer simply needs to have cameras on site and the possibility of connecting to the internet.
With so-called vSaaS, a big trade-off is ongoing data storage costs. Especially in relation to the one-time expense of an NVR product. Additionally, cloud costs are more predictable. At least in relation to the expense of maintaining on-site storage capacity. After all, the user does not have to deal with failed components.
However, you will have problems if the internet goes down. After all, the cameras will not be able to capture the images. But for that, there are hybrid systems. After all, they use some on-site storage to avoid this eventuality.
But there is another big advantage to cloud-based surveillance. Additional AI intelligence can be added as a service to existing provisions.