When it comes to delivering true flexibility in terms of surveillance applications, the preferred approach is typically the use of a VMS package, which then requires the specification of a robust and appropriately specified server. While off-the-shelf units were once the only choice, today optimised servers are widely available from a number of specialist companies. Here, with the assistance of professional security server suppliers, Benchmark assesses the important considerations.
The Right Specification
Robin Hughes, Sales Director, Secure Logiq
The main question is why are we using servers in the first place? For a long time the basis of most enterprise IP surveillance projects has been the utilisation of many cameras on a sophisticated VMS, using a high powered computer to replace multiple DVR/NVR units.
A specialised server with the correct software can record several hundred HD cameras at a good frame rate for several months. What this means is that in a well-designed enterprise system, correct server selection can offer massive advantages in terms of system design and efficiency, power consumption and rack space.
Optimised servers for HD surveillance are available now that can process in excess of 4000Mbps of camera traffic with over half a petabyte (PB) of on-board useable storage in just 4U of rack space. This is equivalent to 15 COTS (commercial off-the-shelf) servers or the highest specification NVRs on the market. In such a case, the power savings alone could pay for the optimised server in just a few years.
The transition from analogue to IP surveillance has been incredibly slow, mainly due to the difference in skill sets between the security integrator and IT infrastructure specialists.
Software and hardware are also completely different disciplines in the IT world, which is why to date most VMS suppliers are content with recommending or rebadging COTS servers rather than a specialist optimised product.
However, often an IT-centric rebadged server simply doesn’t fit the bill. If pushed, many VMS providers prefer to recommend an optimised solution.
When selecting a server the three important factors to understand: throughput, storage and resilience.
In general, industry calculators for throughput are poor and often give wildly varied results and partial information. When calculating server requirements, there are two important numbers you need for any H.264 device: average bandwidth and peak bandwidth.
Average bandwidth is the sum of the individual I and P frames in one second during routine operation, measured in Mbps (Megabits per second). This number can also be used for storage calculations. Often absent in industry calculators, however, is the peak bandwidth value. This is the maximum potential bandwidth output required should all pixels be changing at the same time.
The server selected must be able to handle the peaks caused by high pixel change on all cameras simultaneously or there is a risk of server failure potentially at the most critical time. On many cameras and VMSs, a bandwidth cap is applied allowing for an easy calculation. However, as a rule of thumb, peak bandwidth could be four times that of a quiet scene and double that of a busy scene.
As an example of the importance of accurate calculations, consider a hotel. Generally there will be a few people walking in the corridors at any one time, a few at the bar and a few in reception. Most cameras will be ticking over on low bandwidth as pixel change is minimal.
Imagine now that there is a fire incident in the middle of the night. Alarms are activated, emergency lighting switches on, accompanied by smoke, flames, sprinklers and general chaos. Suddenly all of the cameras will experience very high motion simultaneously.
As this is the type of mission critical event a system will have been installed for in the first place, the server must be able to cope with the throughput demands. The last thing needed is the core of your system failing and shutting down when it is needed most.
So with bandwidth calculations completed, what are the options? Most rebadged generalist servers are designed for multi-tasking: hosting websites, databases, data back-up, email, social media, etc., that are used on a day-to-day basis. Such servers will also run a VMS.
However, it is worth remembering there is nothing more data-intensive than video, other than multiple streams of HD video. Because of their generalist nature, IT-centric servers will often have a restricted throughput, limiting the number of HD streams recording to a single server.
There is a common misconception that to increase the throughput of a server the processor must be upgraded. However, with specialist knowledge and components, there are better ways to optimise throughput. There are a few server manufacturers emerging who understand how IP video works, who have taken the time to analyse the intricacies and modus operandi of the various VMS options and who are creating surveillance-optimised servers to meet the market demands of increasing camera resolutions without having to make compromises on system performance.
The VMS used and other parallel video applications such as mobile gateway, LPR or video analytics will all impact the overall performance of a server, often requiring multiple servers or virtualisation to achieve the end requirement.
Industry calculators will only deliver an average throughput and storage specification, while also sometimes suggesting the processor that might be required. However, the results will never provide enough information for a server manufacturer to build an optimised solution for the application. Consulting an HD surveillance server specialist can often save considerable time, money, rack space and power, not to mention the peace of mind that a hardware solution has been designed with an in-depth knowledge of both IT and IP video.
Calculating storage correctly is important and asking a server supplier to do the calculations is often the best way to get the optimum results; most specialists in this field would prefer to do this and will guarantee the calculations. Alternatively find a reliable and proven calculator which errs on the side of caution. Most customers won’t have an issue if there is more storage than required, but if there is not enough they will be less than impressed. Adding storage retrospectively can be complicated and expensive.
Specialist HD surveillance servers are available with around 720TB storage before RAID in just a 4U rack. With this many disks, higher levels of RAID should be employed so expect to lose 5 or 6 disks to RAID60 or similar, leaving around 660TB after RAID is applied.
This is not the end of the story in terms of calculating actual useable storage. Other losses that need to be accounted for include operating system and application software (specialist server suppliers will separate these off to separate SSDs in RAID 1 for more efficient operation), internal disk firmware (file system, boot data, shadow files and formatting) and Base10 vs Base2 losses (the operating system calculates using binary and HDDs using decimal).
This means that, for example, an 8TB disk will show as 7.45TB in Windows). As a rule of thumb, allow 10 per cent total losses on the RAW storage after RAID has been applied.
If you are buying a a single server to replace multiple conventional products, think about the repercussions if the unit fails. Most specialist servers built for mission critical applications have multiple resilience levels built in as standard, and feature true enterprise-class components which have significantly lower failure rates compared with normal commodity components.
The most common component to fail in a server is the hard drive. Surveillance drives are a step up from commodity drives. However, most HDD manufacturers will state that if the data is valuable, Enterprise drives should be deployed. RAID5 should be the minimum HDD resilience considered for Enterprise applications. Adding an extra redundant disk in the form of RAID5 plus a hot spare or RAID6 buys additional resilience in the event of HDD failure.
Higher storage solutions should employ multiple RAID arrays (such as RAID50 or RAID60 with hot spares) which increases performance, reduces RAID rebuild time and also reduces the statistical probability that a drive failure can ‘break’ the storage array.
Dual redundant power supplies should be specified as standard and not an optional extra. Whilst reliability and efficiency on PSUs have increased in recent times, this is the second most likely component to fail and will render the entire system useless unless a back-up is in place.
To preserve the integrity of the video archive, drives that run the operating system and applications such as VMS should be separate and mirrored.
Another very important part of the server is the quality and resilience of the RAID card, which is where the hard work takes place. Having a battery back-up for the RAID card will ensure at least the archive is protected and shuts down correctly in the event of power failure.
Where possible, select servers with hot-swap redundant components. This means that failed components can be replaced by someone with little or no IT knowledge without powering the system down.
The server is often the single most expensive piece of equipment in an enterprise surveillance system. A well designed server optimised for surveillance can save time and money. Cutting corners could cost money and reputational damage in the long run.
If installers and integrators dictate exact requirements to a server supplier they could be opening a can of worms. It is best to place the onus on them to recommend a guaranteed solution.
There are some basic questions that should be asked. Does the server supplier have a video department? Are they familiar with the intricacies and infrastructure of the selected VMS? Do they have contacts at, and work closely with, the VMS supplier? Can they carry out calculations independently? If the supplier is presented with a system design (x cameras, y resolution, z frame rate and v archive) and the specified VMS, can they design a solution?
If they can’t do it, or if they won’t guarantee the solution, the question is whether they are the best partner to offer support if things start to go wrong.
Until recently, security industry resources have been dedicated to making higher resolution cameras and supporting software, with commercial off-the-shelf servers being selected to complete the system. The rate of change in IT is exponential compared to the security industry, and those that previously migrated to IP surveillance will have found that historically compromises have had to be made. Simply replacing a three year old server with a specialist unit optimised for HD surveillance can transform an existing system. Compression can be reduced, improving image quality, frame rates can be increased, as can archive periods, and power consumption, cooling requirements and rack space can all be reduced.
As a final thought, recording HD video is only part of the battle. Consideration should also be given to the requirements to decode and display multiple streams of HD live and playback images, especially given the recent popularity of panoramic cameras.
Casper Wu, Product Manager, Surveon Technology
There are many considerations to take when selecting a suitable server for a surveillance system. These include storage options, system architecture and the system application.
Always ascertain whether the server supports storage expansion SAS JBOD, iSCSI RAID, NAS, etc.. Many projects only require a small number of cameras with low recording needs during the early stages, but selected servers should offer a chance to enlarge the scale of the system with more cameras and longer data retention periods.
In simple projects, all-in-one servers can be good enough as the local display and all the functions can be set up and operated easily through one unit. On the other hand, a client-server rack-mount unit could be a better choice for applications with a control room, such as medium and large applications. In such architectures, the set-up and daily operation are carried out through the client machine, while the server is used for archiving purpose.
Depending on the application, the server may need to have certain features catering to specific solutions. For retail applications, it’s useful to have POS or people counting integrated into the system itself. Servers for transportation solutions should conform to vehicle-based standards. For mission critical projects, it’s vital to ensure recording is reliable and robust. In such sites a system with failover functionality is required.
Failover solutions must be flexible in terms of system architecture to implement mutual back-up for multiple units. The failover server not only supports archiving, but must also provide full operational control.
Chris Huan, Senior Technical Account Manager, BCDVideo
So you’ve selected your server, so the next step is to consider multicasting and an appropriate switch. Security installers and integrators demand the best surveillance-based networking solutions, but how can they differentiate the good from the underperforming? Identifying these differences is critical when setting up a multicast network. Things can quickly become overcomplicated, resulting in project delays and poor performance. Fortunately, there are three key factors installers and integrators can look for when deploying a multicast network: simplicity, support and reliability.
Multicast networking can quickly become too complex for security installers and integrators to handle by themselves. When this happens, time and money is spent hiring a specialised networking engineer to clean up an overly complicated installation.
Identifying simple, quick solutions capable of supporting multicast networks will bring greater return on investment to end-users and shorten on-site installation time to mere hours.
Modern processes like Shortest Path Bridging (SPB) eliminate network engineers completely by allowing installers and integrators to simply plug-and-play without needing to learn complicated code.
Even the best networking switches can experience issues. However, not every switch comes with around-the-clock support. Both end-users and integrators want to identify and solve problems as quickly as possible. No one should consider a multicast solution without next-day on-site support. Such a move could severely damage an installer’s or integrator’s reputation with the end-user and lead to guesswork-based repairs.
Along with the necessary support, select a reliable multicast device with a history of reliable performance optimised for video surveillance applications. There are several key indicators including flash reconvergence upon network failure to eliminate outages, improved camera call-up, PTZ control, load balancing distribution, robust failover in software and hardware, single protocol usage, etc..
When multicast networking has these features, a synergy is created between any VMS paired with an SPB multicast-capable switch, since the software only uses one protocol instead of multiple protocols like many traditional network platforms.