Surveillance installations come with a variety of storage options. These range from traditional DVRs to NVRs – or even PC-based for small installations – and server-based storage solutions for larger requirements. In this issue we look at the hard drives designed for surveillance and why they are a better choice than ordinary drives when dealing with 24x7 video storage solutions.
In this article, we look at the larger solutions where storage is not a simple choice. Large projects need specifically designed storage that is able to handle many video streams writing information to the disks at the same time, as well as overwriting older video files once their storage duration is reached. And it needs to do this without losing frames or being overwhelmed with endless streams of video data asking to be saved.
In addition, these systems need to be set up to allow for various configurations in support of reliability and availability, as well as hard drive failure (which does happen). The trick is to be able to handle failure without needing to shut down the system while a drive is replaced. And then there is the requirement to scale. You may only need a certain amount of storage today, but what happens when the rules change and you need to retain your video files for longer, or you increase the number of cameras on your site?
So what are the storage options for high-end surveillance projects? While there are many players in the high-end market, Hi-Tech Security Solutions spoke to three companies that specialise in storage solutions for the surveillance market. The companies offer a similar solution in as far as they sell high-end storage, but the way their products are designed is different, and all have the intention to meet the storage needs of large surveillance projects.
Local storage is lekker
Capsule Technologies, a South African company, has designed its own storage servers to meet the needs of the surveillance market. Instead of using traditional IT storage systems, Franck Martinaux, CTO of Capsule Technologies, explains that Capsule has combined all the storage components into a single box, which can be scaled to meet the requirements of the largest of projects. And while it uses standard IT components for its storage, networking and so on, it has developed its own software to take advantage of the hardware and deliver above average performance, even when compared to the best storage systems globally.
“The primary idea of Capsule Technologies is to simplify the IT part of a project by reducing the amount of servers and storage arrays required to do the same job,” says Martinaux. Determining the server/storage combination in a particular job depends on a number of factors. These include:
• The number of cameras. Each camera writes to the storage and to the database of the VMS.
• The recording bandwidth required, which changes as cameras get upgraded to higher resolutions and new features are added.
• The complexity of the processing due to functions such as analytics, facial recognition, LPR, etc.
“With CAPS-OS, our in-house firmware, we leverage modern hardware with an adapted approach of server and storage visualisation to get more out of servers.”
He provides the following example in comparison to Bosch’s BVMS. Classic 16-bay storage servers will be guaranteed, in most cases, for 128 channels and a recording bandwidth of 475 Mbps. A Capsule 16-bay solution will guarantee 400 channels and a recording bandwidth of 740 Mbps because of the optimisations Capsule has implemented. In the South African context, 400 channels qualifies as a large site.
But there are different configurations users can choose depending on their requirements. Martinaux says larger sites can be provided with a single building block (server) that can handle more than 600 channels at 1100 Mbps, and then scale the number of drives in the server to accommodate the required retention period.
For this type of performance, Capsule offers the following products:
• ICE-24: 4U, 24 drives and up to 240 TB of raw storage.
• DR4860: 4U, 48 drives and up to 480 TB of raw storage.
• DR9060: 4U, 90 drives and up to 900 TB of raw storage.
“By combining our building blocks, we can obviously scale up in performance and the storage capacity. For very large surveillance project with 5000, 10 000 or even 15 000 cameras, such as a smart city project, Capsule offers its own CAPS-OS as a software defined infrastructure component (SDI) to be deployed on existing servers in a data centre.
“Using CAPS-OS as an SDI will optimise the workflow and reduce the number of required servers and storage. CAPS-OS is the first SDI developed specifically for the surveillance market,” notes Martinaux.
Different requirements
Of course, a storage solution is not only dependent on the storage servers, users also have to choose a video management system (VMS) as a platform to control their video footage, and potentially many other factors (like access control, alarms etc.). Capsule is certified to work with Bosch, Milestone, Exacq and Axxon at the moment, with more certifications in the pipeline.
Martinaux explains that certification is important as each VMS has a different way of interacting with the storage and the I/O pattern is therefore different between different vendors. Capsule has developed a re-configurable RAID engine, part of CAPS-OS, to ensure the solution maintains its performance by accommodating the I/O profile of the mentioned VMS vendors.
From a SAN to a NAS
While it is true that traditional IT-based storage systems are not ideal for surveillance, that doesn’t mean that the IT companies are not in the game. Hayden Sadler, regional sales manager for unstructured data in the SADC region for Dell EMC, explains that Dell EMC also offers solutions targeted at the surveillance market.
Dell EMC serves the surveillance market starting with smaller solutions in the form of storage appliances and increases its offerings as the demands of the project at hand grow. In mid-range surveillance projects, a traditional SAN (storage area network) platform would meet the needs of users with RAID options for reliability and resilience. (A SAN is a network that provides access to consolidated block level storage providing easy access to operating systems.)
When one gets to the larger installations, a SAN may not be up to the job. This is where a NAS (network-attached storage) comes into the picture. A NAS is a storage server connected to a network that provides many clients with access to a single storage pool.
Sadler explains that a NAS takes much of the complexity of setting up and configuring SANs out of the picture and offers administrators access to their VMS of choice and a large pool of storage, without having to configure RAID settings or access to different arrays.
Dell EMC produces the Isilon NAS solution for this purpose. Sadler says the Isilon is a scale-out, node-based NAS solution that allows users to reach a maximum of 68 PB (petabytes) of storage, which equates to 144 nodes in a single system. Each node has two controllers to ensure that even with a node loaded to full disk capacity there won’t be any delays or bottlenecks in processing data.
As one’s storage requirements grow, Isilon can be expanded by adding more nodes. Each node, according to Sadler, is self contained and includes its own CPU, cache, disks and controllers. So adding nodes to the storage array does not slow the system down, in fact it can improve the performance as it scales.
However, while adding nodes increases the storage available to the user, all the storage in the nodes is accessed via a single file system, meaning the user doesn’t have to worry about configuration or pointing cameras at different storage servers. The file systems ensures it all appears as one pool of storage. The file system also takes care of securing the storage for users, according to their stipulations in terms of downtime, availability and so forth.
The Isilon range was bought by EMC about 10 years ago (EMC was bought by Dell about a year ago) and was designed specifically to cater to media files – or large unstructured data. This is, of course, exactly what video is and Sadler says this is why the Isilon systems are ideally suited to handling multiple video streams simultaneously.
No defrag, no frame loss
The final solution we consider in this article is Rasilient, a storage solution designed specifically for surveillance. Rasilient is distributed in South Africa by IP Video Solutions (IPVS).
Marnix de Lorm, director of IPVS explains that the founders of Rasilient understood that the random access required by traditional IT systems was not optimal for surveillance as video streams are sequential in nature. In other words, normal applications write bits of data to various positions on a hard disk and it doesn’t matter if there is a delay. If your email takes a second, or five seconds longer to arrive, or if opening a spreadsheet takes a second or two longer it doesn’t matter.
The same approach applies to writing data to disk: spreading a traditional data file over the disk in random available places is fine. This results in disk fragmentation which slows down the performance of a disk over time, but normal IT users can run an application to defragment their disk – which takes a bit of time, but this is fine as it can be run after hours.
When the same happens with video streams we have a problem. The first time one writes to disk there may be lots of sequential files and not much jumping around the disk. As files are overwritten, however, the operating system fits the files in where it can and fragmentation occurs rapidly. This adds to the time it takes to write a video file to disk and to read it from the disk during playback.
When one has a number of cameras continually streaming video to the storage system, the disk will eventually get so busy that some data will be lost, which is known as frame dropping or frame loss. The result is you may end up with the video missing frames here and there, and Murphy’s Law dictates it will be the frame with the perfect image of an intruder’s face – or something as important. And with real-time surveillance, you can’t go back and get the footage again.
One can defragment your surveillance disks regularly to reduce this problem, but this means that your surveillance stops for however many hours the process takes. The more storage you have, the longer it will take. And if you do it while running your surveillance operation, even more footage will be lost as defragmentation is quite a disk-intensive job.
De Lorm says Rasilient has developed and patented a number of technologies to ensure that it can guarantee your storage won’t drop any frames and is always available. This intelligence as well as Rasilient’s caching solution is built into the storage arrays to ensure the 24x7 performance of your disks.
Rasilient caches video files and then writes a long sequential file to the disk. One of the patents Rasilient owns is a way to ensure fragmentation doesn’t happen, so the disk will be filled with long uninterrupted blocks of video data. This not only makes reading and writing the data faster, it also places less strain on the disk’s components and reduces heating as there is less work for them to do. Rasilient’s claim to fame is that it will not drop a single frame, and when it’s time to overwrite older data, it follows the same process to ensure fragmentation does not happen.
Users can set up the storage arrays as they require, opting for various RAID configurations, as well as availability and redundancy options. The system also has a backup battery which ensures that if the power goes out, the data left in the caches is written to the drive as soon as power is restored.
Of critical importance to the Rasilient solution, according to De Lorm, are the tools Rasilient offers that provide in-depth insight into what is happening on their storage servers and disks. Users are able to set the system to raise an alert if a single frame is dropped, while also detecting and managing the workload on the system as well as individual disks.
Rasilient uses Seagate surveillance drives in its servers and storage as De Lorm says these have the lowest failure rate. However, the solution also has its own software to determine the health of its individual drives, with self-healing technologies capable of dealing with most problems. When a drive is faulty or something is wrong (such as only one of the dual power supplies is working), users are alerted and can take action without causing any slowdown or shutdown.
The main storage head unit of a Rasilient solution can support up to 12 hard drives with a capacity of up to 120 TB. Further JBODS can be added with 12 drives per device and up to 8 additional JBODs to be able scale up to more than 1 PB per controller. Rasilient offer dual controller options and can support up to 2 Gbps per storage head unit. Each controller has a mirrored cache ensuring data protection should one controller fail.
The three storage brands mentioned in this article are not the only ones available to the surveillance market. Capsule, Dell EMC and Rasilient all offer solutions tailored to large projects that require significant storage resources, the assurance that the data recorded from a camera will be there when it is needed, but also the added benefit of being able to manage the storage with the least hassles.
For more information contact:
Capsule Technologies: www.capsule-sa.co.za
Dell EMC: www.emc.com/en-us/storage/isilon/index.htm
Rasilient: www.ipvs.co.za
© Technews Publishing (Pty) Ltd. | All Rights Reserved.
printer friendly version