VMS Remote Monitoring Tested

What about testing other VMSes? Since this question comes up for every test, let me pre-empt it with our priority for remote monitoring tests.

We will test other VMS remote monitoring capabilities but the priority will be on systems that support transcoding (e.g., Ocularis X, 3xLogic Aztech, etc.). If you know of a VMS that has a sophisticated transcoding capability, recommend it, otherwise no. Thanks.

As a Genetec dealer I love reports like this... It's nothing we've run into before (thankfully); but from what I know of their developers and product managers, they are very sensitive to feedback like this, especially from IPVM (as I am sure most manufacturers are), and I think it will only result in a vastly improved 3.2 WebClient.

Does IPVM have plans on testing Smartphone based mobile clients as well?

This is a timely article for me because one of our customers had asked me to investigate the interactions of Milestone Clients (Smart and Web) with the Server he bought from us.

We were doing performance testing on our Workstation machines so it was easy to do this investigation.

In our lab we established the utility Server to draw streams from and used either SmartClient on a workstation or WebClient on a laptop.

For MS, the key is where the transcoding is performed and what can trigger it.

For SmartClient, the transcoding trigger happens when one sets up the image quality in the Properties for a camera. The documentation says "While using a reduced image quality helps limit bandwidth use, it will—due to the need for re-encoding images—use additional resources on the surveillance system server."

The net on that is to keep it on "Full" quality or get a LOT more CPU in your system to service the transcoding.

For Web Clients the transcoding is always performed by the WebServer application. Do not make the mistake of installing this on your recording server as this application will always transcode every single camera to MJPEG before sending it out to the WebClient.

The net here is to run this app on a dedicated system with lots of horsepower and plenty of network capability.

We also pulled the WebClient up on Smartphones...which also uses the Mobile Server. The 'App' that MS provides does a good job of this.

An overall design consideration we took away from this is to consider the use of cameras that can Dual Stream where you can establish the H264 stream to record with and a MJPEG stream for the live streams. The MJPEG live streams use significantly less CPU to transcode and allows a Server to not get overwhelmed feeding Clients. The caveat here is you need to design in the network bandwidth.

Just wondering if you have tried multi-site VMS performance, not just streaming from one site? The reason I ask is because I am working on a project for a client that needs multiple sites (about 10-12 2MP cams per site) fed back to HQ via broadband based networks. Right now it is 4 off-site locations, but sure to grow to more in the near future.

We have been testing the Network Optics HD Witness program for this scenario and have found some pros and cons. The thick client works pretty well for our test setup of a few cams, we are limited by the 4 camera trial license.

However, the web client will only show streams from cameras that are local to the Enterprise Controller, unless a VPN is enabled (Hamachi). We have found that the VPN is limiting our bandwidth somewhat, but don't have a measurement yet. We would like to avoid using the VPN because of the overhead it adds.

We are working with Netwok Optics at the present time trying to get this worked out. Also, we noticed that in the 2.0 version we could override the settings for camera resolution and frame rate, but those settings are lost once the Media Server is restarted. After a restart of the Media Server software, HD Witness takes over the streams again and you lose your customizations of resolution and frame rate.

This is important to us because we plan on using a 3MP camera, but only need to record 2MP (1080p). We are only chosing the 3MP camera for other features not found in the 2MP models. We don't need the extra resolution and the disk space it would waste.

I have had a different experience with multi-streaming than reported here.

We recently did an Exacq/Axis install that included a number of 5 megapixel cameras on the exterior, and all M30 series on the interior. The internet connection is terrible. The 5MP recordings were unusable remotely, and the 1/2/3MP M30 streams were not a whole lot better.

I configured all cameras with secondary CIF streams at 4fps, and moderate to high compression. I set Exacq to record both streams. This wastes space and CPU on the server - but with resolutions so low it's negligible to the point of being unnoticeable.

I was very surprised by how well it solved the problem. Response was speedy, and you can do a lot more investigation at CIF resolutions than you would think.

Salient touts dynamic resolution scaling which automatically changes resolution before transmission. Would be very interested to see how they stack up considering there high emphasis on this topic.

Salient White Paper

As I understand you have only limited the bandwith in this case.

We regularly see different problems with transmission links that is not necesarily bandwith related. Delay on the links can be significant and also packet errors.

With high latency links Web clients can meet severe problems due to the inherent structure of the HTTP protocol.

So other issues than limited bandwidth can give significant impact on your remote operation.

• While multi-streaming with a secondary low stream is billed to improve remote monitoring, in our tests, many VMS's low stream took almost as long to load as the high stream with negligible benefits in real-time performance.

Any ideas why this would be the case?

John and group, I've made some 'interesting' discoveries regarding remote sites as well, that may be worth sharing.

I most often use Milestone Essential for single site applications for under 20-30 cameras, as I feel its 'thin client' viewing interface (Smart Client) is the most user friendly (esp. in viewing recordings, although I still prefer the old 6.0d version), and the cost is most attractive in this price sensitive market.

However, I have a growing customer base that have multiple remote sites, with 2-5 cameras at each remote site. And, employees at these remote sites are becoming more engaged in using the cameras locally, for local operations support - so video stream access at the remote site is now becoming a requirement as well. This will become a key issue in solution architecture.

Also, users at 'corporate' are not very excited about viewing cameras at these remote sites in less than full HD (1 MP, typically) - after all, that's what they bought- high res. You can see the challenges mounting.

Initially, I tried the obvious - H.264 of course, and high compression and very low frame rates. I've found that even 1 FPS is OK to most users I'm serving for recording and viewing, as long as its HD, for these remote sites.

Here's where I 'discovered' a very interesting 'feature' of Milestone's essential/Enterprise platform - in that when recording at very low frame rates in H.264, Milestone's software 'overrides' the GOV rate (or key frame refresh rate) with it's own settings. For example, at 1 FPS, Milestone will set at GOV/KFRR of 1 keyframe per second - essentially, JPEG! You can imagine this is not very useful for minimizing data rates on a typical T! (1.5Mbps) connection. (To be fair, I undertand their Corporate platform allows you to set this rate, but its too expensive for this customer base)

I've tested 4 other VMS's for this low frame rate/remote site performance (a subset of the ones you've tested here) , and most will use the camera's settings for GOV/KFRR. So, much better for viewing and recording remote cameras.

However, as user's at the remote sites get more engaged with using the cameras locally, you now have video streams going both ways, essentailly doubling the bandwidth requirements, and quickly using too much of a T1 for video, and impacting local applications - including remote terminal apps for centralized servers. A problem...

I've found that the best solution is not the cheapest, and has several counterveilling drawbacks, and that is to put a local recording server at each remote site. The drawbacks are mostly obvious - expense, maintenance, phyical security - but balancing the performance expectations at both HQ and locally is very challenging, with limited bandwidth - almost always a T1 - and not impacting local applicaitons that also must use the connection.

With, this architecture, I've found I like exacq's capability to limit the total amount of bandwidth the server 'feeds' to connections off the local network, combined with their thin client's viewing capabiliites to view multiple servers simultaneously. This means that I can have several viewers remote from the local site viewing the cameras and drawing video streams, an not worry about using more than the 'allotted' bandwidth on the connection for video. And a veiwing client can see the cameras at multiple sites in a single instantiation. It appears that bandwidth limitations are mitigated in the software by reducing frame transmission - OK in most instances.

I can also, of course, set up a second stream at lower bandwidth and frame rate for recording at an off-site server as well - say at HQ - at the cost of an additional license. This is kind of a bummer, as I don't think you should have to pay twice for a camera license for a paticular camera, just because you're recording it on multiple servers - ought to be a strickly 'per camera' license, in my opinion.

Both Milestone and exacq are great platforms, with different strengths and weaknesses...

Undisclosed, Thanks for sharing your experience, I appreciate it. A few thoughts come to mind:

1. There isn't multi streaming (using multiple, independent streams configured differently) in Essential. There is Live View option that only some cameras support (one stream with different settings for fps, etc) which then would allow to have a different setting for recorded. However, if the camera does not have it lowering fps/res/compression will sacrifice recorded quality, storage, and fps since the same stream settings would apply to both live and recorded.

2. The workaround of creating local servers to band-aid the remote capability of a VMS does cost more as you mentioned due to additional licenses (as well as a myriad of other redundancies), but it also maxes out the number of streams each camera can serve concurrently, overloading it- some servers may not stream, until another concurrent server is no longer using the same camera (some cameras max out at 3 concurrent unique streams for either direct access via browser or multiple servers).

1. Yes, that is indeed a limitaion of Essential, but the real issue I had was the GOV/KFRR issue at low FPS rates....

2. Should have clarified that I'm only recording locally - the option to simultaneously record at a central location is avialable/possible (I use AXIS cameras almost exclusively) but I've not found a good reason (beyond physical security redundancy) to consider it. So. only one steam to server per camera (locally) and one license per camera - but woould be nice if licenses were linked to camera and not server so option was available without additional cost...

BTW, one of the really nice multi-stream options on the same local server (with exacq) is to set a second stream to be a time-lapse - set to record at say, one image an hour. Customers really like this to go back and very quickly get an overall understanding of equipment movements, inventory level changes, etc... and in this instance, the same camera license covers it, since it's on the same server.

I'm newbie in this arena but it seems that these manufacturers could create a side application for remote live viewing delivered via a CDN. Allow the VMS, as an option and possibly an additional revenue stream, to stream (based on how much upstream bandwidth is available at each site) to a CDN. It of course will be slightly buffered but would allow 100s of live view streams and it could eventually be done for event based recordings too. Have some intelligence built into it so it knows you are remote and not on the LAN and charge like AWS does - only on usage. This may only be an option for larger spread out customers. I need to check AWS pricing - it may be outrageous. Hopefully this wasn't a ludicrous post

Is it worth testing?