Onvif profile G cameras - disappointing number of devices

[SilverstarAnalytics 0]

Hi All,

 

The onvif profile G standard was released on the 1st July 2014, which is great news for us application developers.

 

The disappointment is the very slow update from the camera manufacturers. I see very little on google and the onvif certification list only has 9 conformant devices from 3 vendors.

 

I've contacted 5 different vendors about their plans to support this API on their edge cameras and get the 'we don't know' response.

 

I'm looking for a cheapish edge camera that supports onvif profile g. Can anyone help?

 

Cheers

 

Simon, SilverstarAnalytics

[jhonovich 0]

The bigger issue is getting VMSes to support ONVIF Profile G. Even if you had a cheap ONVIF Profile G camera, what do you plan to connect it to?

 

While I am sure uptake of Profile G will be slower than Profile S (for streaming), it is just 3 months since G was officially released. I am sure there will be dozens, if not more than a hundred by early next year.

[SilverstarAnalytics 0]

Thanks. We are an VM's (or better described as Managed Video as a service). With all things cloud, we are keen to have as much in the cloud as possible and have a minimal physical footprint/configuration. Thus our focus on edge recording cameras which are directly connected to our cloud service.

 

For us Onvif profile S has been a positive experience (single piece of code that works seamlessly with a wide range of cameras). The next logical step for us is to remove the final piece of needless complexity and use the on-board storage. That way we have an clean and efficient architecture with the cameras being a simple plug'n'go.

 

I was disappointed after a long, long run-up to release, that the Camera vendors didn't have a part of their offerings ready to go at the release of the G specification and today have no clear position if their currently shipping edge recording cameras will have this support added.

 

Oh well, we stick with the proprietary Axis api until Profile G momentum builds.

 

Cheers

 

Simon, Silverstar Analytics

[jhonovich 0]

I'd recommend that you get your software officially ONVIF G conformant. That way, it will make it easier for the camera vendors to have software to test against.

 

From talking to manufacturers, it's a little bit of a chicken and egg dilemma to get G conformant. Each side needs to test against other conformant products, but those are in short supply.

 

One thing that will help is that Genetec plans to be officially G conformant soon (they already have the code in production and they are on the technical committee). Once, Genetec is official, it will spur many of the camera manufacturers who want to get Genetec to support them. That should help you too.

 

Btw, you are probably already aware of this, but the other big issue with edge storage is that a lot of (low to mid tier) SD cards have problems reliably recording video surveillance.

[buellwinkle 0]

A typical modern day NAND flash storage like an SD card has life of about 100,000 write cycles or 274 writes per day over a year. So depending on activity, it could last a few months or a few years. I don't see this as a practical storage method for surveillance cameras and don't see edge storage including profile G taking off unless theres a breakthrough in SD cards. Not sure the manufacturers care as their primary use is cameras and smartphones.

 

I have the extreme best use case in an indoor camera that gets a few motion detects a month and it's been working for about 3 years now. But outdoors, with constant video motion detects, I doubt it will last a year.

 

Would be cool if they had cameras with very small spinning disk built in like the 160GB 1.8" HDD put in the iPod classics like this one - http://www.ebay.com/itm/NEW-MK1626GCB-1-8-160GB-HDD-FOR-iPod-Classic-6th-Gen-Replace-HS161JQ-Hard-Drive-/200881466817

[40th Floor 0]

Leaving aside yet another flawed analysis of expected time of failure for flash storage, I offer my point of view:

 

I've had many HDs fail. I've also have had many run for as long as they remain useful.

 

I've had zero CFs, SDs, SSDs fail. My CFs go back to 10 MB size. I suppose a CPU is more reliable (never had one of those fail, either), but next to them, flash storage is right up there. I don't think anyone (should) consider spinning rust reliable.

[jhonovich 0]

40th Floor,

 

Axis has acknowledged problems with SD cards used in cameras. In response, Axis tightened their recommendations. Here's a copy from the statement they shared with IPVM:

 

"Axis strongly recommends using SanDisk Extreme cards as these were extensively tested without timeout failure," adding that "Enhanced memory cards like SanDisk Extreme are much better suited for demanding applications, such as professional 24/7 surveillance applications. Lower graded consumer SD cards – mainly used for multimedia applications – will in many cases work fine, but the bigger variation in quality and durability of these cards makes them less suitable."

[SunnyKim 2]

jhonovich,

Does ONVIF provide C like - source code that conforms and describes ONVIF specification?. For example, H.264 provides C source codes conforming to it.

[jhonovich 0]

Sunny, No, I have not seen any ONVIF supplied source code.

 

They do have a test tool that one can test their own ONVIF implementation against.

 

That said, to your broader point, there would certainly be value of them providing reference code to simplify and better standardize implementation.

[SunnyKim 2]

jhonovich,

Thanks for your information. That's a Sin if they already knew a line of code states more clearly than a page of text, eventually spreading the adoption more quickly.

[40th Floor 0]

"Axis strongly recommends... using SanDisk Extreme cards

This year's "extreme" is next year's "normal". You could buy junk SD cards today, sure. I would say the same thing about buying junk HDs. Don't do it.

[jhonovich 0]

40th Floor, your opinion aside, I am relaying numerous significant field events of SD cards being used in production IP cameras.

[40th Floor 0]

The IP camera -will- fail in numerous ways before the SD card in it does, if it does. You are so used to everything else failing in the camera (HW, FW or connecting SW, network, power) that apparently you have come to ignore it.

 

Are all your cameras on a UPS? (I can answer that - mine are.)

 

Anyway, flash is more reliable than the camera. Trying to make an argument the other way is...marketing.

[jhonovich 0]

40th floor, you have provided no evidence to support any of you claims.

 

To the contrary, we have a dozen plus independent reports from Axis end users with problems using SD cards. Indeed, as I have earlier cited, Axis has admitted the problem.

[SilverstarAnalytics 0]

Hi All,

 

Thanks for all the responses and info.

 

jhonovich: Yes, we are considering Onvif membership. I don't know how much sway we will have compared to the build boys. We have been considering joining in the past, to help shape the conformance tests, as today we believe they are limited. We find that cameras that work for the first hour or two, have issues after prolonged testing. We would want to contribute some soak tests to ONVIF. It's surprising but the majority of camera vendors have issues in the area. The one vendor who doesn't in our experience is d-link so hats off to them.

 

buellwinkle, 40th Floor: Yes, the weak point of edge cameras is cheap SD-cards. In our experience (just like with cheap ebay camera specials) cheap no-brand sd-cards are asking for trouble. The higher end SanDisk ones are the items to have. The good news is that Moore's law will sort this out soon. It won't be long before we have 128Gb cards as the norm with the better memory technology and all this early stage FUD will go.

 

SunnyKim: I totally agree with you. The onvif docs and example source code are pitiful. Our engineers gave up on the supplied info and found it quicker to use wireshark running applications.

 

For us, edge recording cameras are a game changer. It allows us to address the devices from the cloud and retrieve the 0.1% of valuable video, whilst leaving the 99.9% of dross on the device. This way, deployment costs are much reduced, network bandwidth usage is minimal and is all round a better way forward.

 

Cheers

 

Simon, Silverstar Analytics

[40th Floor 0]

J- You didn't answer the question. From that I presume you don't. Here is a tip: put any camera that writes to flash on a UPS. Just do it. Before the UPS shuts down, instruct the camera to stop writing. Welcome to Computers 101. When you graduate, then we can talk.

[buellwinkle 0]

40th, does you CastleKeeper support edge recording and the new Onvif G specs, just curious.

 

As for Moore's law, storage is one area where it hasn't applied maybe doubling in performance every 10 years instead of each year like processors. When I quoted 100,000 writes, that's from Sandisk Extreme specs, most cards are rated for 10,000 writes. But I get that it makes sense for cloud as a way to store most recent video before sending it offsite.

[40th Floor 0]

ck2 was done in 2009 and released in 2010. It supports whatever was around back then, even h264[1]. There is a web page devoted to it where you could read about it if you really wanted, Carl. It's very old stuff, though., serviceable as it is (October will be 5 years running here with no crashes).

 

 

It's not like you are into new things, Carl, even if they aren't all that new. Come on, don't pretend you know things you don't.

 

[1] http://www.networkcamerareviews.com/forums/about4008.html

 

Only kidding. Sort of.

 

SD cards are cheap. If you are scared of it going bad out of the blue (as if nothing else will before, and instead), replace it like you would anything on a service schedule, be it at 274 days or at 2 years.

[buellwinkle 0]

Did not know you gave up on CK years ago.

 

Maybe just paranoid, but I'm worried about the NAND flash even in the camera because I have an Arduino with a temp sensor updating the OSD for a camera with temperature readings on a web cam. Just a trait of NAND flash that includes SD cards, SSD and flash chips in cameras. Was going to do it once an hour, but will change it to 2-3 times a day and hope for the best.

 

Maybe I don't know much, but 35 years working as a software engineer and developer, I may have picked up a few things. Actually never thought about SSD's failing sooner than spinning disk because of excessive writes until one of our hardware engineers brought this up as many of the SSDs that we sold a few years ago are starting to exceed their life expectancy and when you have maybe 1,000 of them in one appliance, that can be maintenance headache for sure.

 

As for life expectancy of storage, got a WD Purple 3TB because that's the drive they recommend for NVRs. Well that lasted a week until it died, just got the warranty replacement yesterday.

[40th Floor 0]

ck2[1]: Fork inserted. Done. There was one DLL updated a couple years ago.

 

software engineer and developer [for 35 years]

I call myself a programmer, sometimes computer programmer. Always have. No funny looks with that. I know the "engineer" thing has been around for nearly 25 years, though. "Developer" doesn't say much about anything, either. Anyone in the process could be one (art, design, marketing) so that's kinda like saying "I am a worker". But then you sure had me fooled so what do I know.

 

As for fear of failure: How many times has the power gone out there? How many times has a flash failed? One example of things that go wrong all the time, and for which most do nothing about it (like using a UPS). Write to any drive and while doing so pull the plug. Luck is the all that keeps very bad things from happening (on FAT-based drives, especially).

 

... starting to exceed their life expectancy and when you have maybe 1,000 of them in one appliance, that can be maintenance headache for sure.

Yup. That's why things have service schedules. Replace before their time, or cross your fingers. I heard Intel's SSDs do a Logan's Run after so many writes, but I also heard of SSDs far outlasting their design life.

 

As for the WD purple, a textbook case on how to market a bottom-end drive to those who should know better. What in the world got you to go with that, other than the marketing -- it got you, huh? LOL

 

[1] But the next thing is simply amazing. Well, to me.

[buellwinkle 0]

I differentiate software engineer as someone that work on lower level code, in my case operating systems vs. application code, but done that too. Never in 35 years did I ever have the title of programmer, I would have done something else rather than endure that, LOL. But outgrew that, no longer satisfied with "engineer" in my title, I've grown to scientist, a notch above an engineer, only because there's a civil engineer that I know that's so incompetent, I refuse to be called an engineer anymore.

 

Marketing hype or not, the purple is the same price as their desktop drives and right now the replacement drive is sitting on my desktop, hence it's now a desktop drive aka. paperweight.

[40th Floor 0]

Programmer fits perfectly for me. That is exactly what I do.

 

Onvif profile G... By the way, and I think you asked back there, I got into this topic because of the flash angle (life of), not flash as storage on a camera specifically. It has its place, but if the network is there, and its fast, I would always store off-camera. I can pull 6 Mbps x 20 over 24 hours, just as a start, without any concern, and access that at any time, instantly. If this were stored on-camera, I'd have very slow access (65 GB a day, per camera). If a camera was made to fall back to on-camera storage when its network was down, that would be good (if only it worked every time). Or if it only recorded on (rare) motion or event, that would work.

 

I saw WD's 16+ ms average access time for its purple and ... beyond belief; that was slow 20 years ago. That's one thing, but to then see places like THG say "[nevermind that slow access because WD tells us so]" is ... what's beyond beyond belief? Critical reviews have left the internet - no biting the hand that feeds.

 

Jumping forward: The hard drive will be gone in 10 years. They know it.