Scalability is a growing focus in today’s video surveillance market. The explosion of new IP Cameras, IoT devices, access control, audio devices and more is driving intense new scalability requirements for video surveillance systems. And the mainstreaming of Full HD resolution (1080P) and 15fps and higher recording is being exacerbated by a seemingly endless march towards longer and longer mandated retention times.
Even smaller sized security systems that start out with 30 to 50 cameras are growing by leaps and bounds in terms of camera count, video motion processing and video retention. Going from a single all-in one appliance with 20 to 200 cameras to a distributed system that comprises many hundreds of cameras at exponentially higher bandwidth and processing rates is not an easy task.
Scaling up a Server vs Scaling out A Network
Infrastructure scalability handles the changing needs of an application by statically adding or removing resources to meet changing application demands as needed. In most cases, this is handled by scaling up (vertical scaling) and/or scaling out (horizontal scaling).
Scale-up is done by adding more resources to an existing system to reach a desired state of performance. For example, a database or web server needs additional resources to continue performance at a certain level. More compute, memory, storage, or network can be added to that system to keep the performance at desired levels.
Scale-out is usually, but not always, associated with distributed architectures. Scaling out is essentially adding additional infrastructure capacity in pre-packaged blocks of infrastructure or nodes. Think multiplying appliances rather than enlarging a single monolithic server.
Both approaches work well and have various pro’s and con’s. Video Surveillance, however, has its own unique challenges and data flow architecture that makes it lean towards scaling out appliances rather than attempting to cram it all into one monster box.
All of the leading global Video Management System (VMS) companies have somewhat similar system design architecture. All of them have some sort of database performance or capacity limit (aka the limits of free database apps). All VMS’s attempt to be flexible and open and allow customers to choose the resolution, bandwidth, fps and analytics of each video stream. This means that there are wildly varying performance implications for every single video surveillance installation. And then you add in growth and a flexible, scale-out architecture works best.
Arxys engineered the VideoX NVRs to be high performance, hardware accelerated building blocks to enable you to grow your video surveillance installation without disruption.
Using a scale-out system design you can scale your video surveillance system seamlessly and easily. Add more cameras, more devices, higher resolutions and frame rates, ensure every video stream is running video motion detection with analytics metadata and empower smart search across every video stream. Scale out NVR’s to add cameras and retention.
The other benefit of a distributed, scale-out topology is that taking advantage of the advanced failover features of Milestone Corporate and Expert, ONSSI Enterprise and other top end VMS versions delivers improved security and availability. And it does it at much lower cost than traditional scale-up designs.
In video surveillance the best practices are to use a pool of Failover Servers to serve a group of active Recording Servers. (See above) To improve reliability the pool should have N+1 Failover Servers, with N= the highest number of planned concurrent outages due to regular maintenance. The Failover Servers are sized to support which ever active Recording Server in the group is the largest, therefore any Failover can easily take over for and support the full load of any active Recording Server from the group.
Take full advantage of Arxys VideoX tremendously lower costs per camera and activated hardware acceleration by using a network appliance building block approach. Scale-out to support many thousands of cameras and devices.