jump to navigation

I’m sorry I think you’re wrong…and heres why July 1, 2009

Posted by fermorm in Uncategorized.
Tags:
5 comments

Well, he would say that. wouldn’t he?

 “Everything that can be invented has been invented.”

–Charles H. Duell, Commissioner, U.S. Office of Patents, 1889

 I read with interest the article last month by Mike Newton of Dedicated Micros and I have to say that my first thought was one of disappointment, and my second; “Well, he would say that wouldn’t he?” I was disappointed that Mike seemed to use the chance to open a really good debate on the future of IP vs. Analogue CCTV as a promotional piece for Dedicated Micros, and also to try and hold back the market in an unhappy compromise of hybrid systems.

 The things people say…

 “I think there is a world market for maybe five computers.”

–Thomas Watson, Chairman of IBM, 1943

I also had a sense of déjà vu when I read Mike’s article. In my career I have often heard similar arguments. When I started in the IP field in Data Capture and EPOS (Why would you ever want live data? – Batch processing is always best) through to general IT (What? – Connect every computer in the world to every other one? – Are you mad?), then to Wireless Communications (Why on earth would I ever want to log onto my email or work remotely from a coffee shop?) before coming to rest at present in IP CCTV (No, NVR’s will never take off – VCR Tape will always be best). At every step there have, and always will be, those who either cannot or do not wish to see the future but yet tomorrow always becomes today and those who stand still will be left behind.

Speaking as the Country Manager of a Value Added Distributor for IP CCTV solutions and with my experience across a number of IP fields I feel that I should at least try to put the other side of the IP vs. Analogue debate. There are many independent market research reports that state that at some time in the near future IP will replace analogue based solutions as the technology of choice. Even Mike acknowledged that IP is the future, but I think his pitch was slightly skewed by the fact that Dedicated Micros don’t have a pure end-to-end IP solution,and indeed no vendor does have the complete end-to-end solution, as it’s not the current business model of the market: – Everyone plays to their own strengths be they recording, camera and / or encoder manufacturing or storage / transmission.

However, as the entire physical security market of Intruder,Access Control, CCTV and even potentially Fire, moves to an IP based platform, both System Integrators and End Users will request and expect totally integrated IP solutions. From the customer’s perspective, there are three main deployment scenarios: (1) integrating existing analogue edge components with an IP backbone, (2) migrating existing analogue installations to IP over the life cycle of the system and finally, (3) installing a complete Greenfield pure IP solution that meets the project needs, budget and allows for future proofing.

All these scenarios present challenges in both planning and deployment, but these can be identified early enough to ensure that there are no surprises and the project can be designed, managed and delivered within agreed timescales and within budget.As explained below, Mike’s estimated £1.8M cost for a 750 camera system is wrong and misleading. I would love for a client to be prepared to spend £1.8M on a system, but I doubt such a customer readily exists. I think that Mike has been overtly disingenuous with some of the statements he has made regarding the cost of deployment for his theoretical 750 camera system; and I do not agree with the cost analysis, ROI arguments or system supply figures that Mike has quoted. Unsurprisingly though, we’re in agreement that the end user should never face any hidden costs or the need to deploy a “costly, high speed backbone”. But that’s never been an IP-related issue in my experience.

Remember the 5 p’s – proper planning prevents…

 “There is no reason anyone would want a computer in their home.”

–Ken Olson, President, Chairman and Founder of Digital Equipment Corp, 1977

The secret to successful IP solution deployment lies in acknowledging two simple factors: (1) fully understanding the concept and requirements of system design for IP including how everything fits together and (2) how and where to source all the correct components from multiple manufacturers.

When looking at any of the three deployment scenarios above there are different challenges to each with regard to existing system components, so it’s worth looking at any additionally required components and responding to some of Mike’s comments. It should be noted that through good system design each project model can be deployed in a number of ways – central storage, distributed storage, redundancy, failover and remote high quality recording with low quality live view are all possible. These solutions can easily be implemented across existing infrastructure (without impacting IT services such as email, IP telephony or data processing applications), or new IP CCTV deployments with the use of networking functionality such as VLAN’s (the ability to create separate “virtual” LANs on the same switch) or by utilising low cost switches to create a separate LAN for the surveillance system.

However, I struggle to understand Mike’s concept of requiring 211 servers to run video analytics. Video analytics was one of the most over-hyped features of IP based video post 9-11. It was to be the Nirvana of video solutions allowing us to spot and identify suspects’ faces, track suspicious behaviour and generally replace the need for CCTV operators. Sadly this is not yet the case. That said, however, video analytics has made great strides forward and does have a key role to play in the future of IP CCTV.

Today it is possible to analyse video for incorrect motion or direction, loitering, left or removed objects and a host of other functions very accurately. In the past this did require a low camera feed to server ratio due to the limitations of the hardware platforms however with the improvements in processor and memory technology this argument no longer holds water as most IP camera manufacturers are looking to deploy analytics on the cameras or encoders themselves to reduce this server loading.  

Some of the technologies Mike references such as Transcoding are available to a number of vendors within the IP or NVR space so this should not be seen as a unique feature of a Dedicated Micros based solution but there are a number of features available to IP solutions that are not available to hybrid solutions such as intelligent maps, integration with other systems such as Access Control or multi site co-locations.

Building the perfect system

“I used to think that cyberspace was fifty years away. What I thought was fifty years away, was only ten years away. And what I thought was ten years away… it was already here. I just wasn’t aware of it yet.”

–Bruce Sterling, Science and Technology Author, 2000

In order to dispute Mike’s assertion that a 750 IP based camera system would cost £1.8m compared to Dedicated Micros £250k or £600k solution, I have to make a few assumptions concerning the system proposed by Mike. Mike hasn’t provided a detailed system design in his cost comparison model, so we’ve put forward a high-end solution proposal as follows:

·          750 camera feeds

·          l31 days retention of footage and a 10% additional archive for incident recordings.

·          All streams to be recorded 24/7/365 at 4CIF / D1 at 25fps

·          Recordings are of average motion of 50% in view atany one time. (Why is this of interest? Different encoding technologies mean that significant savings can be made by only sending changes in the frame rather than a complete refresh every frame. H.264 encoding can offer between 50% and 80% savings over M-JPEG for instance.)

·          All cameras are already installed and running via coax to local collection points and then high speed CAT5e or CAT6 cabling needs to be installed back to a central command and control point. There is a difficulty here in that we have no “real” site to work from so the assumption will be made that the cameras are grouped in groups of 16 and are no more than 300m from the central control point on standard runs through existing conduit that would be there for the existing infrastructure or cameras (they all need power, right?)

·          As the system is truly mission critical 100% redundancy and failover are required.

Using a leading vendors freely available storage calculator on the above figures the following storage profile is generated (for this example Axis Communications tool was used so readers can check these figures for themselves).

This design tool shows that only 116.2 TB of storage is needed – even when adding the additional 10% overhead for incident recordings (or approx 16 TB), the storage total is only approx 135 TB. It’s possible to purchase 16 x 1 TB iSCSI drive units which would mean that you would only need 11 raid arrays running in RAID 5 (I’ll cover these terms and the benefits in a moment).

Then when we look at the required servers to process and record the cameras, due to increases in processor power and memory, this means you can support on an average COTS (Computers Off The Shelf manufacturer such as Dell or HP) platform 64 camera feeds therefore only requiring 12 servers plus an additional one to handle system redundancy, not the 24 servers required in Mike’s solution.

The use of iSCSI (Internet Small Computer System Interface) means that the storage arrays are connected directly to the network backbone thus eliminating the potential for a single point of failure in the traditional DVR route. Should a drive lose connectivity, power or suffer a systems failure then the network control software automatically sees this and starts to record its streams to another array in the stack seamlessly.

By coupling  this with RAID (Redundant Array of Inexpensive Discs) even in the event of a drive failure the data is duplicated across multiple discs so that no footage is lost  So to connect all this up we simply need a good qualityswitch. A Layer 3 or Layer 4 managed 10/100/1000 switch (Layer 3 or 4 refer to the level within the OSI (Open Systems Interconnection) stack model to determine where the switch manages network traffic.

Layer 3 being the network or IP address layer and Layer 4 being the transport or port layer. Think of an IP address as a phone number to an office and then the ports as being your unique extension – If you don’t give the port number to someone directly then they have to go through your receptionist and they may not allow certain calls through).

With the above server count and storage arrays you simply need 2 x 48 port switches and the cabling to connect all your remote collection sites. With the cost per port dropping on switches and even with high grade cabling all this could be installed for less than £25k. Even if you had to deploy this over wider areas then other IP technologies, such as secure wireless, still make this a cost effective solution.

This covers our central management and recording hardware. Mike makes no mention of operator stations. Perhaps he assumes to continue to use the old analogue control room? But again to gain the benefits of a true IP solution this could all be easy, quickly and inexpensively upgraded to 60” progressive scan HD screens supporting 16:9 widescreen resolution thus improving image quality, detail and operator experience.

The most expensive part of an IP CCTV solution will always be the camera licensing, but as this normally is around only £150 per camera (Which is often for life – If you hardware fails you just reinstall.) I don’t see this as a major issue.

The final component is simply to then add encoders at the edge to convert analogue signals to IP. With the new ranges of IP encoders from the leading vendors it is possible to convert these streams and transmit them using H.264 at a bandwidth for 16 cameras of only around 12Mbps per group. Again the cost per channel for this is sub £200 so no high costs there either. To address Mikes promotion of embedded storage on edge devices, while this may seem practical at first, it simply doesn’t work in the real world as there is no way to insert this footage back into your central recording software / device due to the need for watermarking and audit trails.

Furthermore if you design the network  correctly you can build in redundancy through network protocols such as Spanning Tree and OSPF (Open Shortest Path First) to ensure that there is no single point of failure in the network. The use of protocols such as these actually would give you even more availability than a traditional analogue based solution when coupled with UPS (Uninterruptible Power Supplies) and POE (Power over Ethernet). So even before we discuss other ROI benefits of IP solutions such as POE (reduced cabling costs and installation time), Intelligent deployments (Dynamic storage profiles, redundancy and integration with other hardware systems), or Intelligent Video (analytics integration with other software systems or improved user functionality), we can see that IP is the future of CCTV

I have seen the future and IP works

“The Web as I envisaged it, we have not seen it yet.The future is still so much bigger than the past.”

—Sir Tim Berners-Lee, Computer Scientist and credited with creating the World Wide Web

Using the above specification, the cost of this hypothetical solution – the recording servers, the software, the encoders, the transmission, display equipment and installation and commissioning for a high frame rate IP solution for 750 cameras – comes in at an End User price of less than £500k excluding VAT. The cost would be even lower at 5fps as the storage would be quartered and costs would be reduced on transmission and encoder technology with more cameras supported per server.

The launch of organisations such ONVIF to facilitate open standards in network video, the industry names committed to it, and independent market research from organisations such as JP Morgan, IMS, and Frost & Sullivan showing IP will overtake analogue as the technology of choice should only confirm that the future lies in IP and digital technology.

Just as it did with computing, mobile phones, telephony and television, IP will revolutionise the CCTV industry. The challenge for us all is to adapt to this rapidly changing market and ensure that our customers are able to utilise the advantages IP offers.

 The future is here now and IP works. For more information about Controlware visit www.controlware.co.uk

Advertisements

Taking A Closer Look At The True Cost Of Centralised Nvr Based Cctv June 23, 2009

Posted by fermorm in Uncategorized.
4 comments

We believe users should take a close look at the real, vs. imagined, merits of the differing approaches to CCTV in an IP environment. In recent years security and IT managers across the globe have been bombarded by enticing claims from vendors regarding the ROI (Return On Investment) that they could expect to realise if they chose to adopt a centralised Network Video Recorder (NVR) based CCTV solution. On the surface the inflated claims of some IP camera/NVR (Network Video Recorder) proponents, left unchallenged, have invariably played a key role in influencing buying decisions. The reality, we would contend, for those swayed by such arguments has not always been as straightforward as the widely promoted figures may at first have suggested.

Uncovering the Infrastructure Challenges In fact with IP-only systems there are often hidden costs associated with the need for an expensive network upgrade to cope with the new demands placed on an organisation’s IP infrastructure. Without such enhancements many solutions struggle to reach anywhere near the quoted performance levels. On the ground some end users have become so disillusioned with the whole process that they are even contemplating replacing installed IP cameras with traditional analogue models to overcome their ongoing network problems.

Alongside this, given the current economic conditions, the preferred or practical route for end users is not necessarily going to be the wholesale replacement of a large legacy population of analogue cameras and control equipment with solely IP systems. Rather we suspect, moving forward, what is likely to prove more realistic, deliverable, and offer a better long-term ROI is a hybrid route for CCTV by allowing systems, both analogue and digital, to grow seamlessly in line with the end user’s future surveillance needs.

 When it comes to the recording, transmission and storage of CCTV in the IP environment we have consistently advocated a distributed or decentralised server model, with Dedicated Micros latest DV-IP Server being a case in point.

Distributed Surveillance

It needs to be made clear that we are not against IP CCTV in fact, DV-IP actually stands for Distributed Video over IP and DM have been providing IP Video solutions for over 10 years. Rather the rationale behind our approach is driven by a number of key considerations which ultimately, we feel, deliver a high level of performance, whilst minimising set-up and infrastructure costs and, crucially, provides an all important safety net in the event of unexpected network failure.

We do not advocate, for instance, that end users should be placed in a position where they automatically need to invest in a costly, high speed, backbone so their network can cope with the new CCTV infrastructure. As a consequence AD Group NetVu Connected solutions – including those from Dedicated Micros – come with features such as TransCoding technology which allow users to capture, view and playback high definition images on the local LAN whilst simultaneously streaming bandwidth optimised images to remote users without impacting the network.

Additionally by taking a distributed, de-centralised approach – with multi-node architecture – where recording actually takes place at the edge of the network on the hard drive of a unit such as with the DV-IP Server it is possible, in most cases, to support fail safe continuous recording with on demand access to all parts of the system. This is a critical advantage compared to pure IP/NVR solutions which are totally reliant on the status of the network for their ongoing operation. Any network downtime will, consequently, have a detrimental impact on their ability to function and to capture all important digital video evidence.

 This is an opportune moment to take a more detailed look at the IP/NVR (Network Video Recorder) concept and compare and contrast this, in terms of potential costs, with what Dedicated Micros currently offers with regards to a distributed, embedded, alternative.

Challenging the IP/NVR Agenda

When it comes to the evolution of the IP camera market, the element which has played a key role in its evolution is the NVR (Network Video Recorder). This has been developed very much from an IT rather than security centric position, however it is a flawed model for a 24 hour failsafe environment. The IP Camera/NVR model is totally dependent on the integrity of the network for recording, viewing and playback.

Although the costs of the IP cameras themselves have been coming down in recent times to around $520 (£361) per camera – according to JP Freeman – there is the cost of a typical software license for a NVR/Video Management System (VMS) to be considered which can be from £10k to £50k for systems of between 50 and 750 cameras. Where the large-scale costs really hit home is when we look at a typical high performance IT server capable of handling 32 channels of video at5 fps (frames per second).

Beyond this, should analytics be required as part of the surveillance solution, then realistically another server is going to be needed for every four channels of video – each channel must be decoded in realtime within each server before the analytics can be performed.

In fact, in a recent system tender for an IP/NVR based solution by a leading supplier, the individual licence costs per camera, plus the per camera share of the failover server costs exceeded the complete embedded server hardware costs, without separate licences, and then of course the servers, network attached storage and all the other hidden costs as well still to be added.

So back to the potential costs of the example above, looking at a 750 camera system may require around 24 servers – if analytics are excluded – and a maximum of 211 if it is required. This is only the tip of the proverbial iceberg. There are also high network integrity requirements to be considered, alongside the need for storage above and beyond that provided through the servers, so this has to be accommodated through other elements such as Network Attached Storage. The upshot of all this is, not surprisingly, a potential capital requirement of approximately £1.8 million and that is only at 5 fps (frames per second).

This is in contrast to what, we are confident, can be achieved in an IP or hybrid model distributed to our new DV-IP Servers. At less than £5k per server, and 48 servers, the project could be delivered for a basic cost below £250k. If full frame rate DI recording performance is required then using DV-IP RT 16 channel units the base equipment cost is still compares favourably at approximately £600k

The lesson for security and IT managers is to be wary of vendors who claim that they can provide a dream IP/NVR solution for just the £50k license fee, as we have shown above in extreme cases this could result in equipment costs of nearly £1.8 million simply to provide the infrastructure to support the required performance levels. Placed in context this certainly makes our hybrid DV-IP solution at c£250k an extremely attractive option and for those looking for even higher performance with real time DI images and the DV-IP HD for under £600k is certainly a realistic proposition. There is a growing realisation, as the above figures demonstrate, that the network dependency and real system costs of a pure IP/NVR solution mean it is simply not a commercially viable option in many scenarios.

Another pitfall for a software only NVR/IP approach is its ability – or rather inability – to deal with a need to display multiple full frame rate D1 images together. Realistically a high specification IT server is typically only capable of displaying between four and eight MPEG-4 streams and as little as four AVC streams.

This compares to a solution we can offer with an embedded decoder, inside our distributed servers, each giving 48 full rate D1 streams to two HDMI displays, or 16 full D1 AVC decodes.

Also there is a popular misconception that video walls can only be achieved with the PC/NVR environment on multiple PC’s into a video wall controller, the reality is that embedded multi-screen decoders offer far more powerful, robust cost effective solutions.

A New Approach with the ICR

So where do we go from here? Well the position we have held at Dedicated Micros over many years is that analogue and hybrid analogue/IP distributed servers offer a cost effective and reliable solution. We still feel that this holds true, particularly given the cost of ownership issues highlighted above.

Having said the above we also recognise that there are some specifiers and end users out there who, what ever we say, are fundamentally committed to an IP only model. They may see even our latest IP enabled solutions, such as the DV-IP HD, as still reminiscent of previous DVR-based solutions – although we certainly beg to differ on that point. As a consequence there is a growing momentum in the marketplace, we believe, for a distributed or decentralised model based around an IP camera which actually pulls together all of the key elements of effective CCTV in one single unit. This is something have actively pursued and can now offer the Integral Camera Recorder (ICR).

The Integral Camera Recorder (ICR) which we have developed at Dedicated Micros comprises a full, enterprise grade, server – similar to those in our DV-IP range – integrated within an IP camera, ensuring that recording can be maintained regardless of the IP status of a network, this fits in with our distributed or de-centralised approach to CCTV recording. In terms of the storage of CCTV images by the ICR we hope to be able to achieve this in three ways. Firstly through internal storage which could take the form of one or more SD/CF cards, typically providing a maximum rate and resolution for one or more days.

As well as the internal solid state storage, by microSD card or similar, the ICR can also utilise further tiered storage, including a locally connected SATA drive over ATA Over Ethernet (AoE) which will continue to operate, even in the event of complete failure of the TCP/IP architecture and infrastructure network level failures. The images can also be centrally recorded and archived in parallel in the same manner as an NVR based solution – but with all the high availability network risks completely mitigated.