Industry Experts Chart the Future of Video Storage Solutions
1
S With video surveillance systems now being mainstreamed into IT in many enterprise organizations, how is this feature-rich technology going to reap the business value given its potential for advanced analytics and integration of video with other systems and data elements? ecurity Technology Executive editorial director Steve Lasky recently queried two of the industry’s top video storage and surveillance technologists to investigate the future. This continued series of technology- sponsored Roundtables-in-Print includes Amit Gattani, who is currently Senior Director, Segment Marketing for Embedded Business Unit for Micron and is actively involved in driving Micron’s strategy and business in various IoT segments, along with Ken Mills, the Global Industry IoT, Surveillance and Computer Vision General Manager for Dell Technologies Corporation, who is regarded as one of the company’s top surveillance and public safety experts. STE: As we look to the future of video surveillance storage solutions what do you see as some of the most dynamic and innovative drivers that bring new usage models and value to this technology sector? Amit Gattani -- Advances in artificial intelligence, machine learning and deep learning techniques for video analytics, along with the rapid rise in availability of higher-resolution cameras at cost- effective prices, is driving next-wave innovation in the video surveillance industry. There is a growing use case of very large-scale deployments with richer datasets that need to be inferred in real- time to drive better business intelligence and outcomes. From a storage solutions perspective, this drives demand for higher-performance storage, driving a shift from traditional hard-drive storage to flash-based storage systems on the infrastructure side, and demand for more edge storage on the camera side. Ken Mills -- We are seeing many of our customers accelerate their adoption of cloud technology models, “multi-clouds” that balance an optimal mix of public, private and edge cloud resources for their businesses to deliver a consistent experience with common management. With this framework in mind, we will continue to see video surveillance solutions that integrate with cloud models direct to cloud storage, scale-out storage (on-premise /hybrid/ to cloud) along with expanded functionality (machine learning, artificial intelligence, API’s, etc.) coming from continuous integration and continuous deployment capabilities. With this faster processing availability, CPU, GPU, FPGA and TPU, cameras will also gain more capability to actively participate in these dynamic models as well as create a truly intelligent edge. STE: Edge storage devices have been around for quite some time, but with advancements in technology, the business case is evolving. Discuss how increased memory retention and the ability to move storage to the camera itself can create new and more effective surveillance architecture solutions and what end-users should be looking for with regards to seamless video surveillance management and their storage infrastructure? Gattani -- In the past 20 years, on- camera edge storage in surveillance has primarily been used as a network backup. With limited use cases, most memory solutions were generally low-density, low-retention, and low-performance in nature. The emergence of higher resolution cameras and the demand for powerful real-time analytics are the trends demanding more processing and storage at the edge/endpoint, to reduce the burden of bringing and storing a lot of raw data in the cloud or centralized servers. On the other hand, NAND technology has been advancing rapidly with 3D vertical stacking, and up to four bits per cell design (QLC), enabling us to create ultra-high- density in-camera storage (microSD form factor) solutions in capacities where primary storage can now be completely in the camera. This delivers a much lower Total Cost of Ownership (TCO) than most systems in existence today. NAND original manufacturers can also fine-tune manufacturing processes and the firmware to deliver a storage product with higher endurance, no video frame drops, and health management software that make primary storage on the camera a competitive reality. Combine this product capability with the surveillance market trends mentioned above and we are beginning to drive a significant design shift towards an advanced analytics-rich camera with built-in primary storage. Regarding what designers should be looking for as they architect primary storage solutions on the edge, SD and micro SD are obviously great choices as they deliver the flexibility of a removable solution while delivering the size needed in the camera form factors. Designers need to differentiate between the “type” of solutions, consumer-grade vs. industrial/surveillance grade. If a typical consumer card was inserted into a 24X7 recording environment, it would fail within a few months, as those cards are not designed to sustain the number of NAND write cycles and will wear out. Hence it’s important to select a surveillance-grade product with appropriate endurance and capacity based on weeks of desired data retention, high reliability (like two million hours MTTF), long warranty (3-Year+ warranty) and a surveillance optimized feature set like frame drops management and predictive health monitoring of the storage. Mills -- With many physical security providers migrating toward more cloud-native models, the capabilities that containerization and micro- services deliver for edge devices such as cameras that deliver significantly more processing capability, larger local storage (including caching) and faster connectivity (802.11ax, 5G, etc.) are dramatically improved. Not only can video be cached and streamed at higher resolutions (4K, 8K, etc.), but also the processing ability allows for significant artificial intelligence outcomes and scale-out services between the cameras to create a considerable force-multiplication capability operating with maximum flexibility and capacity at the edge. Cameras can become a true Fog Operating Platform. STE: Can you provide some of the most compelling use cases for these evolving video storage solutions and what their impact is on improved security and safety? Gattani -- Sure, I’ll give two examples, one centered around cost, and the other performance. 24X7 recording and monitoring is a growing requirement as our industry evolves. The cost of data transmission and storage on the cloud is significant and perpetual. Local on-camera storage reduces that expense, and when combined with cloud management services, it can provide the same level of service required by the end-user, at a cost that allows continued development and innovation of edge services. On the performance side, a good example is threat detection and alert. As image recognition and deep learning algorithms for threat detection mature, the demand to do more comprehensive analytics on richer data sets with lower latency/real-time results will drive the need for computing and storage to be on the edge. Mills -- Offering solutions that can significantly impact the safety and security of a city, community or business while increasing the quality of service delivery always brings a compelling result. Deploying a cloud-native and more importantly, a multi-cloud approach, video surveillance solutions that include cameras, sensors, storage and more will be able to take advantage of the capabilities from both the manufacturer’s platform along with the scale of mega-clouds and niche API players to deliver artificial intelligence/ machine learning outcomes that provide proactive insights. If the storage array is running out of storage, you can immediately scale to cloud. If a safety issue is being detected and there is a specific description or criteria needed to be located, it can be quickly enabled whether from the camera itself or via the VMS. We also suggest leveraging existing cameras (typically the largest part of the cost) and integrating those with a Computer Vision Platform that provides significant capabilities such as AI, Machine Learning, Camera as a Sensor, etc. without having to change out most of the existing infrastructure. STE: VSaaS will continue gaining momentum, especially for small businesses and enterprise customers with multiple small locations. Discuss why this option is fast becoming a cost- effective business decision for video surveillance data storage and eventual data retrieval. Gattani -- I agree that VSaaS is going to continue to gain dominance as the solution of choice for SMB. The reasons why are simple. The need and benefits for surveillance in our world are increasing, the cost of an on-premises install is prohibitive (NVR, networking equipment, VMS software), and the resource to monitor and maintain systems is expensive. In addition, security is not the core competence of these SMB owners, nor do they want to develop it, especially when they can pay a VSaaS provider to do it more cost-effectively than they can do it themselves. With a solid five years of ramping trend and a significant growth forecast in the future, VSaaS suppliers are considering the most cost-effective and efficient storage architecture to offer to their customers. As discussed earlier, advances in non- volatile memory technology have enabled edge models that drive cost and performance advantages that VSaaS suppliers are innovating with, delivering value with, and ultimately differentiating themselves from their competitors with. By leveraging high-density edge storage in the camera, VSaaS providers can reduce the cost of high bandwidth data transfer to the cloud and large cloud storage, delivering a better TCO to the customers while continuing to deliver the benefits of a cloud-managed solution. Mills -- Consumption-based technologies are a natural fit for organizations that find it challenging or ineffective to use capital expenditure models to procure Video Surveillance as a service (VSaaS). The core strategy of VSaaS ensures that the devices will continue to be updated, monitored, and secured rather than static systems that are purchased at a point in time and rarely updated or managed properly. STE: With the growing migration of video to the edge and the expanding IoT universe, what will be the impact on Cloud Services as they relate to video storage and analytics? Will Cloud-based video applications and products continue to take off? Gattani -- With IoT, one of the major shifts occurring is that video surveillance is being integrated into an overall IoT framework for larger deployments like in smart cities, enterprise, campuses, etc. IT managers need one unified solution stack to manage all their IoT deployments, including video surveillance and access control. This is creating an opportunity for a hyper-converged intelligent solution stack in the cloud that allows various sensors and systems to be fused together for more intelligent and complex decisions and productivity enhancements. Video surveillance is no more about just theft management and security, but about enhancing customer insights and improving productivity, be it in retail, manufacturing, transportation, etc. While camera- level analytics is localized to the camera for the reasons already discussed, cloud opportunity is strong for multi-camera, multi-sensor integrated decision making and analytics. Mills -- Cloud Service Providers play an important part in Video Surveillance Solutions for scale, capability and capacity. Depending on the customer’s business models (centralized, distributed, small, large, etc.), Cloud Providers can play a holistic role for a customer or provide a specific value such a burst capacity (compute, storage, etc.) or artificial intelligence outcomes where the Cloud Provider has a unique or better capability than the native platform being used. STE: Some video solutions vendors are using cybersecurity protection to differentiate their equipment from that of their competitors. How are OEM storage vendors partnering with camera manufacturers to create a secure ecosystem for the vast amount of video data stored on a user’s network? Gattani -- Cybersecurity is critical in video surveillance and all IoT applications and must be considered end-to-end. It starts with ensuring that all end devices, be it video cameras, access controllers or other smart sensors can be “trusted” – can be identified and authenticated by the network on an ongoing basis. This requires the use of hardware roots of trust embedded with the flash or working with the flash in the camera where the camera code is stored, to ensure active measurement and attestation of the code and resiliency in case of a potential hack. This is even more critical in the era of machine learning usage in the camera, as a hack of the neural network parameters can completely change analytics outcomes. Micron’s Authenta solution is an example of that. These same roots of trusts can be used to create certificates for establishing secure connections to protect all the data communication back to the cloud. Local edge storage cards must provide access control protection for data, and systems may encrypt the data per application need. Similarly, on the network/cloud side, the servers need to have their own hardware security to prevent against hacks to the code itself, and data also needs to be encrypted. Memory and storage OEMs like Micron have a comprehensive set of built-in security in our products, best-practices knowledge, and vast ecosystem support to help our customers design for end-to-end comprehensive security right from the start. Mills -- Although there are many advances occurring in the broader video surveillance solution industry, we are just in the initial stages of seeing deep integration between cybersecurity and physical security. Video surveillance solution providers still have an opportunity to further integrate with Identity Access Management Integrations including Federated Security, Vulnerability Assessment Platforms, Secure Software-defined Networking, Endpoint Detection and Response, Encryption (In-Flight and At-Rest) as well as many other valuable integrations. These integrations ensure that only the right users have access from the right devices/locations, the platform itself is constantly validated to protect from any known vulnerabilities with any software component of the platform, along with secure configuration, optimal network design and secure network configuration and that the platform can defend itself based on intent. Amit Gattani Ken Mills Continue Reading LEARN MORE ABOUT Dell & Micron ROUNDTABLE Industry Experts Chart the Future of Video Storage Solutions ! " # $ % & ∠
Transcript of Industry Experts Chart the Future of Video Storage Solutions
S
With video surveillance systems nowbeing mainstreamed into IT in manyenterprise organizations, how is thisfeature-rich technology going to reap thebusiness value given its potential foradvanced analytics and integration ofvideo with other systems and dataelements?
ecurity Technology Executive editorialdirector Steve Lasky recently queriedtwo of the industry’s top video storage
and surveillance technologists to investigate thefuture. This continued series of technology-sponsored Roundtables-in-Print includes AmitGattani, who is currently Senior Director, SegmentMarketing for Embedded Business Unit forMicron and is actively involved in driving Micron’sstrategy and business in various IoT segments,along with Ken Mills, the Global Industry IoT,Surveillance and Computer Vision GeneralManager for Dell Technologies Corporation, whois regarded as one of the company’s topsurveillance and public safety experts.
STE: As we look to the future ofvideo surveillance storagesolutions what do you see assome of the most dynamic andinnovative drivers that bring newusage models and value to thistechnology sector?Amit Gattani -- Advances in artificial intelligence,machine learning and deep learning techniquesfor video analytics, along with the rapid rise inavailability of higher-resolution cameras at cost-effective prices, is driving next-wave innovation inthe video surveillance industry. There is a growinguse case of very large-scale deployments withricher datasets that need to be inferred in real-time to drive better business intelligence andoutcomes. From a storage solutions perspective,this drives demand for higher-performancestorage, driving a shift from traditional hard-drivestorage to flash-based storage systems on theinfrastructure side, and demand for more edgestorage on the camera side.
Ken Mills -- We are seeing many of ourcustomers accelerate their adoption of cloudtechnology models, “multi-clouds” that balance anoptimal mix of public, private and edge cloudresources for their businesses to deliver aconsistent experience with commonmanagement. With this framework in mind, wewill continue to see video surveillance solutionsthat integrate with cloud models direct to cloudstorage, scale-out storage (on-premise /hybrid/ tocloud) along with expanded functionality(machine learning, artificial intelligence, API’s, etc.)coming from continuous integration andcontinuous deployment capabilities. With thisfaster processing availability, CPU, GPU, FPGA andTPU, cameras will also gain more capability toactively participate in these dynamic models aswell as create a truly intelligent edge.
STE: Edge storage devices havebeen around for quite sometime, but with advancements intechnology, the business case isevolving. Discuss how increasedmemory retention and the abilityto move storage to the cameraitself can create new and moreeffective surveillancearchitecture solutions and whatend-users should be looking forwith regards to seamless videosurveillance management andtheir storage infrastructure?Gattani -- In thepast 20 years, on-camera edge storagein surveillance hasprimarily been usedas a network backup.With limited usecases, most memorysolutions weregenerally low-density,low-retention, andlow-performance in nature. The emergence ofhigher resolution cameras and the demand forpowerful real-time analytics are the trendsdemanding more processing and storage at theedge/endpoint, to reduce the burden of bringingand storing a lot of raw data in the cloud orcentralized servers. On the other hand, NANDtechnology has been advancing rapidly with 3Dvertical stacking, and up to four bits per celldesign (QLC), enabling us to create ultra-high-density in-camera storage (microSD form factor)solutions in capacities where primary storage cannow be completely in the camera. This delivers amuch lower Total Cost of Ownership (TCO) thanmost systems in existence today. NAND originalmanufacturers can also fine-tune manufacturingprocesses and the firmware to deliver a storageproduct with higher endurance, no video framedrops, and health management software thatmake primary storage on the camera acompetitive reality. Combine this productcapability with the surveillance market trendsmentioned above and we are beginning to drive asignificant design shift towards an advancedanalytics-rich camera with built-in primarystorage.
Regarding what designers should be looking foras they architect primary storage solutions on theedge, SD and micro SD are obviously greatchoices as they deliver the flexibility of aremovable solution while delivering the sizeneeded in the camera form factors. Designersneed to differentiate between the “type” ofsolutions, consumer-grade vs.industrial/surveillance grade. If a typical consumercard was inserted into a 24X7 recordingenvironment, it would fail within a few months, asthose cards are not designed to sustain thenumber of NAND write cycles and will wear out.Hence it’s important to select a surveillance-gradeproduct with appropriate endurance and capacitybased on weeks of desired data retention, highreliability (like two million hours MTTF), longwarranty (3-Year+ warranty) and a surveillanceoptimized feature set like frame dropsmanagement and predictive health monitoring ofthe storage.
Mills -- With many physical security providersmigrating toward more cloud-native models, thecapabilities that containerization and micro-services deliver for edge devices such as camerasthat deliver significantly more processingcapability, larger local storage (including caching)and faster connectivity (802.11ax, 5G, etc.) aredramatically improved. Not only can video becached and streamed at higher resolutions (4K,8K, etc.), but also the processing ability allows forsignificant artificial intelligence outcomes andscale-out services between the cameras to createa considerable force-multiplication capabilityoperating with maximum flexibility and capacity atthe edge. Cameras can become a true FogOperating Platform.
STE: Can you provide some ofthe most compelling use casesfor these evolving video storagesolutions and what their impactis on improved security andsafety?Gattani -- Sure, I’ll give two examples, onecentered around cost, and the otherperformance.
24X7 recording and monitoring is a growingrequirement as our industry evolves. The cost ofdata transmission and storage on the cloud issignificant and perpetual. Local on-camerastorage reduces that expense, and whencombined with cloud management services, it canprovide the same level of service required by theend-user, at a cost that allows continueddevelopment and innovation of edge services.
On the performance side, a good example isthreat detection and alert. As image recognitionand deep learning algorithms for threat detectionmature, the demand to do more comprehensiveanalytics on richer data sets with lowerlatency/real-time results will drive the need forcomputing and storage to be on the edge.
Mills -- Offeringsolutions that cansignificantly impactthe safety andsecurity of a city,community orbusiness whileincreasing the qualityof service deliveryalways brings acompelling result.Deploying a cloud-native and more importantly, amulti-cloud approach, video surveillance solutionsthat include cameras, sensors, storage and morewill be able to take advantage of the capabilitiesfrom both the manufacturer’s platform along withthe scale of mega-clouds and niche API players todeliver artificial intelligence/ machine learningoutcomes that provide proactive insights. If thestorage array is running out of storage, you canimmediately scale to cloud. If a safety issue isbeing detected and there is a specific descriptionor criteria needed to be located, it can be quicklyenabled whether from the camera itself or via theVMS. We also suggest leveraging existing cameras(typically the largest part of the cost) andintegrating those with a Computer Vision Platformthat provides significant capabilities such as AI,Machine Learning, Camera as a Sensor, etc.without having to change out most of the existinginfrastructure.
STE: VSaaS will continue gainingmomentum, especially for smallbusinesses and enterprisecustomers with multiple smalllocations. Discuss why thisoption is fast becoming a cost-effective business decision forvideo surveillance data storageand eventual data retrieval.Gattani -- I agree that VSaaS is going to continueto gain dominance as the solution of choice forSMB. The reasons why are simple. The need andbenefits for surveillance in our world areincreasing, the cost of an on-premises install isprohibitive (NVR, networking equipment, VMSsoftware), and the resource to monitor andmaintain systems is expensive. In addition,security is not the core competence of these SMBowners, nor do they want to develop it, especiallywhen they can pay a VSaaS provider to do it morecost-effectively than they can do it themselves.
With a solid five years of ramping trend and asignificant growth forecast in the future, VSaaSsuppliers are considering the most cost-effectiveand efficient storage architecture to offer to theircustomers. As discussed earlier, advances in non-volatile memory technology have enabled edgemodels that drive cost and performanceadvantages that VSaaS suppliers are innovatingwith, delivering value with, and ultimatelydifferentiating themselves from their competitorswith. By leveraging high-density edge storage inthe camera, VSaaS providers can reduce the costof high bandwidth data transfer to the cloud andlarge cloud storage, delivering a better TCO to thecustomers while continuing to deliver the benefitsof a cloud-managed solution.
Mills -- Consumption-based technologies are anatural fit for organizations that find it challengingor ineffective to use capital expenditure modelsto procure Video Surveillance as a service (VSaaS).The core strategy of VSaaS ensures that thedevices will continue to be updated, monitored,and secured rather than static systems that arepurchased at a point in time and rarely updatedor managed properly.
STE: With the growing migrationof video to the edge and theexpanding IoT universe, what willbe the impact on Cloud Servicesas they relate to video storageand analytics? Will Cloud-basedvideo applications and productscontinue to take off?Gattani -- With IoT, one of the major shiftsoccurring is that video surveillance is beingintegrated into an overall IoT framework for largerdeployments like in smart cities, enterprise,campuses, etc. IT managers need one unifiedsolution stack to manage all their IoTdeployments, including video surveillance andaccess control. This is creating an opportunity fora hyper-converged intelligent solution stack in thecloud that allows various sensors and systems tobe fused together for more intelligent andcomplex decisions and productivityenhancements. Video surveillance is no moreabout just theft management and security, butabout enhancing customer insights andimproving productivity, be it in retail,manufacturing, transportation, etc. While camera-level analytics is localized to the camera for thereasons already discussed, cloud opportunity isstrong for multi-camera, multi-sensor integrateddecision making and analytics.
Mills -- Cloud Service Providers play an importantpart in Video Surveillance Solutions for scale,capability and capacity. Depending on thecustomer’s business models (centralized,distributed, small, large, etc.), Cloud Providers canplay a holistic role for a customer or provide aspecific value such a burst capacity (compute,storage, etc.) or artificial intelligence outcomeswhere the Cloud Provider has a unique or bettercapability than the native platform being used.
STE: Some video solutionsvendors are using cybersecurityprotection to differentiate theirequipment from that of theircompetitors. How are OEMstorage vendors partnering withcamera manufacturers to createa secure ecosystem for the vastamount of video data stored ona user’s network?Gattani -- Cybersecurity is critical in videosurveillance and all IoT applications and must beconsidered end-to-end. It starts with ensuringthat all end devices, be it video cameras, accesscontrollers or other smart sensors can be“trusted” – can be identified and authenticated bythe network on an ongoing basis. This requiresthe use of hardware roots of trust embeddedwith the flash or working with the flash in thecamera where the camera code is stored, toensure active measurement and attestation ofthe code and resiliency in case of a potentialhack. This is even more critical in the era ofmachine learning usage in the camera, as a hackof the neural network parameters can completelychange analytics outcomes. Micron’s Authentasolution is an example of that.
These same roots of trusts can be used to createcertificates for establishing secure connections toprotect all the data communication back to thecloud. Local edge storage cards must provideaccess control protection for data, and systemsmay encrypt the data per application need.Similarly, on the network/cloud side, the serversneed to have their own hardware security toprevent against hacks to the code itself, and dataalso needs to be encrypted. Memory and storageOEMs like Micron have a comprehensive set ofbuilt-in security in our products, best-practicesknowledge, and vast ecosystem support to helpour customers design for end-to-endcomprehensive security right from the start.
Mills -- Although there are many advancesoccurring in the broader video surveillancesolution industry, we are just in the initial stagesof seeing deep integration between cybersecurityand physical security. Video surveillance solutionproviders still have an opportunity to furtherintegrate with Identity Access ManagementIntegrations including Federated Security,Vulnerability Assessment Platforms, SecureSoftware-defined Networking, Endpoint Detectionand Response, Encryption (In-Flight and At-Rest)as well as many other valuable integrations.
These integrations ensure that only the rightusers have access from the rightdevices/locations, the platform itself is constantlyvalidated to protect from any knownvulnerabilities with any software component ofthe platform, along with secure configuration,optimal network design and secure networkconfiguration and that the platform can defenditself based on intent.
Amit Gattani
Ken Mills
Continue Reading
LEARN MORE ABOUT
Dell & Micron
ROUNDTABLE
Industry Experts Chartthe Future of VideoStorage Solutions
! " # $
%
&
∠
10/2/19, 6)54 PMPage 1 of 1
