1. From GDPR perspective you cannot store more than 90 days (reffering to e.g. 365 days of retention) - of course there are some exceptions but in most cases you cannot exceed 90 days.
2. What about security of footages stored on SD card? What if I stole the camera with sd-card (od course robbery moment will go to the cloud) but I'm predicting that I can simply restore all data from SD and play them using PC/Mac (same story like typical NVR).
3. Maintenance - as you know some cameras needs special permissions to install them or installation proces can be very complex/hard - what if I need to replace SD card after 2-3 years even in few cameras? From the TCO perspective it can increase the whole investment and make it unprofitable.
4. In hybrid model you need to calculate Internet connection with some additional margin - especially when you are using single provider and number of office users still growing (in some moment can be a bottelneck...).
1) These restrictions don't affect US consumers today. Organizations store for years more often than you would think.
2) Very easy to encrypt an edge storage card. Axis does it with massive success across the entire line minus a few models.
3) Most cards specific to this purpose are rated for 5+ years now. Surveillance technology advances monumentally in 5 years. Worth while to upgrade for technology cost savings far sooner than a camera/modern sd can last.
4) Most cloud platforms can utilize 10-12kb keepalives for health monitoring, etc and edge store evidence. Axis cloud for example can record in massive frame rate/resolution on edge (playable or able to export this from the cloud at any time) and leverage the power of artpec for streaming to the cloud only upon analytics tripping for example and that stream can be a different frame rate/resolution than what is being stored locally. You can get hyper lean on bandwidth with these more modern cloud environments.a
Line 4 - But to elaborate: 87% of our cloud customers leverage very high stream profiles at the edge but also record to the cloud in bandwidth sensitivite methods. Intrusion triggers can start cloud recordings, any type of analytic can trigger that recording, or obvious ways of recording like motion detection, etc to the cloud.
When using an Axis camera in the cloud, we can generate up to 8 different stream profiles that record to different locations with different triggers and different resolution/frame rates, etc.
Our integrators deploy cameras over LTE connections all the time - we easily stay under their bandwidth caps while still hitting surveillance retention goals.
Not sure what platform that is for - "cloud backup" tends to be a poor approach to cloud surveillance in most situations. Having a separate stream entirely directly from the camera that is both live and dedicated to redundancy purposes is a far more effective approach. This stream can then be optimized for the purpose of transit across the internet.
Dear @Jacob Hengel - glad to see that you are start this discussion and your voice is in opposite in my opinion - it's always good to discuss :)
I can only say that I'm still confused regarding LTE/remote/cloud solutions - it's not mean that I'm a "dinosaur" on CCTV area (and only agree for on premise installation). No, no. I like new technologies, new approach to old topics, end so on. But I also a fully aware person and I see obstacles, threatens and "childhood illness" of this solutions.
I have strong prior experience on IT area (networks/OS/security) and this is something which I still missing work in a daily basis with CCTV solutions: every time I need to explain and build awareness regarding possible danger factors.
Of course GDPR is not affect on countries outside EU - you are fully right, but EU market is quite big and we need to also consider this fact.
Moreover example from @Michael Miller showing very clearly that bandwidth utilization is not a isolated case: in my opinion (and based on my experience) it can consume much more than in you examples (of course it's depends on scene but you cannot "cheat" a codecs without loosing quality :)
Regarding HDD vs SD - good that lifetime (MTTF) is still growing but remember about much harder conditions where this card needs to work (usually outside the building, inside the camera - too hot/too cold). It will be cheaper and this is the fact: not tomorrow but soon you will buy 1/2/4TB in a very reasonable price. I only worried about replacement this card in case of failure - as I mentioned: in some cases camera installation process taking days (not hours/minutes) because of complexity, localization or necessary permissions. It's much easier to replace HDD from disk array in server room than replace SD card in above location. Maybe solution can be 2 (two) SD slots cameras (and work in RAID1 settings)? - good question to vendors :)
I believe that next iteration of video codecs and 5G solutions will help us to implement more and more cloud based technologies and for sure this is the convenience and easy way to building CCTV solutions (but again: security, stability, durability).
I am in this industry for the long-term and have a reputation to up-hold; I would not insult anyone with differing opinions :-) Those types of claims are silly and non-professional at best.
In regard to GDPR - more states will continue to enact laws like the CCPA that are built from successes with GDPR. It's simply a matter of time. Not sure how long it will take the US fed to make realistic privacy laws. Responsible organizations like my own however try to target and exceed GDPR regulations because - #1) we work globally but #2) it's the right thing to do and creates trust with our end-users. Transparency is critical.
In regard to storage retention times - that piece has always given me concern. NIST800-53 outlines limiting data to data that must be collected and to not store longer than needed. Ok - vague but a good framework for organizations to take and define what is needed. Now with that being said, statute of limitations in some cases is 24 months or longer. If you have to protect your organization from a slip/fall case that could get raised 23 months after an event - data retention laws being mandated would put commercial businesses in a sticky situation with how malicious lawsuits are here at times.
I love your idea about dual-sd cards. The Axis P3717 for example does have two SD card slots but it is generally used so that 2 sensors go to 1 sd card and 2 sensors go to the other. No hardware/software redundancy of the data there. If 1 SD card failed you could certainly record all 4 sensors to 1 sd card but that doesn't address the data-loss of the first card unless you setup a smart stream to the cloud in the first place.
Technologies like Axis Zipstream on top of h.264 and h.265 can cause massive savings when looking at cloud apps. Very exciting where the world is going.
I am curious as what the 2 million hours MTTF really means. One limitation of flash memory is that, although it can be read or programmed a byte or a word at a time in a random access fashion, it can be erased only a block at a time. In other words, flash memory offers random-access read and programming operations but does not offer arbitrary random-access rewrite or erase operations. Another limitation is that flash memory has a finite number of program – erase cycles (typically written as P/E cycles). Most commercially available flash products are guaranteed to withstand around 100,000 P/E cycles before the wear begins to deteriorate the integrity of the storage.
To understand the life of the unit one needs to know what the P/E cycles limit of the device is, the block size of the erase algorithm, the typical block size of the write and how much free storage is available on the device. The later being important as the device will employ a wear leveling algorithm over all the memory the device to extend life. Spare memory can be used to minimize the writes to the unit. So a 1TB card is really less than 1TB.
SD Card manufactures tend to hide this information, especially the P/E cycles. Micron announced P/E cycles of 1M on some of their devices. It is unclear whether this device has that large of a P/E. If anyone finds these detailed specs please share them. A white paper on how to efficiently extend the life of the SD card would also be extremely valuable.
If this truly equates to a reasonable expectation of MTTF it truly is a breakthrough. If not, who wants to go around and replace possibly 100s of SD cards every year at a installation.
SDcards sound great I'm theory for many applications (remote, few devices,etc) until you thing about the TCO and service impact of relying on a device that will fail but may be challenging to replace. It's very easy to install SDcards while bench configuring a project, but a lot more difficult to roll a truck with ladders/lifts once the cameras are installed.
Also, I've seen some instances where the card has failed but there was no alert for it until missing recorded video was found post-incident. The health status alerting aspect could reside in the camera or software but as an integrator or E.U. I would want assurance that the card is being proactively monitored.
Real cloud solutions deploy not only health monitoring of the camera but these new cards support health monitoring as well as long as the cloud provider is diligent enough to write edge health monitoring into the environment.
Axis has health monitoring in their cloud for 256GB cards, etc. 1TB is hyper-new but these numbers continue to climb on capacity and lower in cost per GB.
I say it’s pretty bold to offer 1 year worth of footage. I know it’s Verderka making the claim but if a client is solely relying on the card as an NVR and it becomes corrupted whose head is on the line, the integrator. I’m all for edge recording because it has help past clients so it will bring competition which is a good thing.
While Verkada is eliminating NVRs, they are not eliminating price gouging, marking up the $279 Micron card for a $2,000+ increase in their 365-day retention models.
Thinking out loud, maybe Verkada expects these cards to fail much more frequently than the SSD drives they currently use? Don't know but they have their 10 year warranty so failures for them could be especially costly over time. But marking something 600% is a good way to cover that.
So Verkada is claiming 356 days of storage for continuous recording of a 5MP video? If my math is right that is around ~300Kbps for 365 days of continuous recording for a 5MP video stream. I don't think they are going to win image quality shootouts at that bit rate.
this 'guarantee' annoys me - with no mention of scene complexity.
I think your math is correct Michael - which means that in order to accomplish the 'guaranteed' claim, they have to use CBR (i.e. your ~300Kbps) - which will certainly look worse as the scene complexity increases.
I disagree. The price isn’t bad. I think 1TB is overkill personally. At least for a single camera. However it would be good in a “HIVE” setting like NX used and allow cameras to failover to each other if on SD card goes bad. We sell Spectrum to homeowners all the time. They end up spending $1,800 to $2,500 or so on a server. For 4-8 cameras. Using SD cards with no server would be an option. I also like that there’s no longer a single point of failure. If a server goes down all cameras go down. If you have just one camera go down all others are still running.
There are 9-figure organizations investing in this technology every year. Many use edge as their only storage target on hundreds of devices.
Many smart cities also leverage edge technology as a redundant storage location with extreme reliability. This use case is different but you would be shocked how reliable the solution is and how often these consumers need to use edge footage because of network or nvr failures.
I tend to agree that we are yet to see field data on the R/W cycles, especially when exposed to outdoor temperature extremes. Not to mention the inevitable "lost in high weeds" cards from those lost / dropped for one reason or another during installation. I look forward to seeing more feedback form the early adopters, as experience is a great teacher.
My biggest take from this article is the following:
"This is ~10x more expensive per-byte than surveillance grade hard drives from WD or Seagate in an NVR, which costs ~$25-$35 per TB";
Can someone share with me where to purchase 1TB WD or Seagate surveillance drives for $35?
From a Law Enforcement perspective, I believe that redundant storage of any kind is a bonus considering thieves occasionally target DVRs/NVRs when facilitating their crime. Additionally, in our current social climate of increased arsons being committed, I believe it would almost irresponsible not to include edge storage when a device has the capability to do so.