Subscriber Discussion

Bandwidth Question - How To Speed Up Pulling Recorded Video From Remote Locations?

NH
Newell Hirata
Jan 03, 2018

I am an end user in a corporate office. I need to access NVRs with Exacqvision VMS at remote locations. When I access the NVR I usually need to pull video from the location back to my office. Our current bandwidth is quite slow. I need 8 minutes to pull 1 minute of video from 2 - 4mp cameras (approx 81 mb). Right now our only solution is to increase the bandwidth between each location and my office. Was just wondering if some of you may have a better solution. Feel free to ask me any questions. As I said I am the end user and not really the tech guy. Thanks for any help and Happy New Year to all.

U
Undisclosed #1
Jan 03, 2018

You don't increase bandwidth 'between' your office and the remote locations....

You can increase the UPLOAD speed at the remote locations (this is the probable limiter), or you can increase the DOWNLOAD speed at your office - or both.

(2)
JH
Jay Hobdy
Jan 03, 2018
IPVMU Certified

Like UD1 said, you probably have slow upload speed at the site.

 

Usually, everyone is concerned with download speed, or how fast can you download stuff from the internet. However, with surveillance, we usually need to upload. 

 

As it turns out, upload speeds are usually much slower than download speeds. At my office, we have around 180Mbps download, and 20Mbps upload on a typical Comcast business account.

 

This ratio is fairly typical (unless you have Google fiber etc) and if you have a cheaper account such as 50Mbps download, you might only have 5Mbps upload which could really slow down video uploads.

 

I suggest having a user at the site do a speed test from a workstation first.

 

More than likely the service plan will have to be upgraded

(2)
(1)
Avatar
Ryan Hulse
Jan 03, 2018

Hi Newell, 

I work for JCI with the exacqVision product.  The feedback from Jay and Undisclosed #1 are correct approaches.  I will also add a few more considerations specific to exacqVision. 

When you say you need to "pull" video, what do you mean by "pull"?  It sounds like you are referring to viewing recorded video or exporting that video.   In those cases, yes, you can't reduce the amount of bandwidth you need, so the most effective solution is to increase the upload bandwidth available on the remote site.   

exacqVision does support setting up multiple streams on a camera, so you could set up a lower quality stream for live viewing which will consume less of the upload bandwidth, but when it comes to exporting the recorded video you can't really reduce it. 

(1)
NH
Newell Hirata
Jan 04, 2018

Thank you for the replies. 

JH
John Honovich
Jan 04, 2018
IPVM

Newell, a few other things to consider:

  • Do any of your cameras support smart codecs and this is not enabled? Or can the firmware be upgraded to support smart codecs? This is likely a long shot but wanted to mention it as this would significantly reduce bandwidth consumption and speed up the transfer.
  • Have you reviewed what compression level your cameras are streaming at? You might be able to reduce the compression, lower bandwidth consumption with no material impact on video quality. For example, many manufacturer default compression levels are, in our analysis, unnecessarily high (see IP Camera Manufacturer Compression Comparison). However, if this is done incorrectly, it will reduce quality so consider carefully.
  • For future camera purchases, I would prioritize making sure they support smart codecs for the same reason as above.

Hope this helps. Any questions, let us know.

(1)
(1)
CR
Chad Rohde
Jan 05, 2018

Didn't see an answer to what "pull" means, but you could use a remote desktop service to access a local PC on NVR network. Only need a meg or two for remote desktop. Maybe you could export on remote PC and do some compressing and then upload, taking a shorter/faster route to where you are if possible.

(1)
Avatar
Simon Lambert
Jan 09, 2018
IPVMU Certified

The amount of data you record (and may need to transfer down a narrow pipe) could be greatly reduced by taking the "2 - 4mp cameras" and reconfiguring to, say, 1 - 2mp or lower. Of course, this is only appropriate if the smaller images still meet your requirements for image detail.

Often people install 1080p cameras (~2 MP) to cover narrow portals such as doorways, gates, etc. for simple identification of people, vehicles, etc. However, this purpose may still be achieved at a 6-foot wide gate using a much smaller 4CIF image (0.4 MP). So by cutting the size of each image to the minimum that meets your purposes you could, in this example, reduce data by 80%. That would download a lot more easily!

Remember that when you cut image resolution by half (e.g. 1080 vertical pixels down to 540 pixels) you cut data requirements down to only a quarter. That's a law of increasing returns.

When folks insist on trying to upsell to larger and larger imagers (way beyond the client's actual need) they don't explain that simply doubling the resolution (distinctly different from image size) will lead to a quadrupling of data. That's a law of diminishing returns to the user, but an incovenient truth to some sales people ;)

(1)
JH
John Honovich
Jan 09, 2018
IPVM

Often people install 1080p cameras (~2 MP) to cover narrow portals such as doorways, gates, etc. for simple identification of people, vehicles, etc. However, this purpose may still be achieved at a 6-foot wide gate using a much smaller 4CIF image (0.4 MP)

Simon, in an age where smart codecs are now the norm in new cameras, you can effectively get traditional 4CIF bandwidth levels while delivering 2MP resolution/details. Are you against using smart codecs or?

A 6' wide area with 4CIF is ~100ppf, which is a lot but 6' wide is quite tight for most applications. Even 12' is on the low side from the calculations we see designers use and then that drops you down to ~50ppf with 4CIF which can start creating noticeable image degradation.

 

(1)
(1)
Avatar
Simon Lambert
Jan 11, 2018
IPVMU Certified

John, my suggestion was meant to as an extra method which would multiply a smart codec's effect. Never meant to be an "either/or" argument. Instead, an "and".

I wonder how my choice of words could have suggested in any way that I might be against smart codecs. You'd already proposed the latter, so I didn't repeat it. Of course, use of smart codecs and configuring smaller images sizes are not techniques in conflict.

Good engineering, as you know, includes designing for efficient use of available resources and within constraints. So lowering the resolution down to simply satisfy each image's purpose is a principle whereby data load can be minimized. I'm sure you will agree that too many installers simply leave their customer's new equipment at factory default when a little thoughtful setting of a few basic parameters would server the owner much better.

(2)
(1)
U
Undisclosed #2
Jan 10, 2018

Show me a 1mp camera. I can show you 10 2mp cameras that offer newer technology and can encode lower resolution profiles, if not full res at lower bitrates - with a negligible price delta. (Probably less expensive given the specific application)

 

JH
John Honovich
Jan 10, 2018
IPVM

Show me a 1mp camera. I can show you 10 2mp cameras that offer newer technology and can encode lower resolution profiles

Not to speak for Simon but I suspect the recommendation would be to buy the 2MP camera (so you get the newer functionalities) but set the stream resolution to 1MP to save bandwidth. As I said above, I think using smart codecs with max resolution is generally better but that would be a third option here.

(1)
Avatar
Jeffrey Hinckley
Jan 09, 2018

I am not sure of your applications and scenes with your cameras, but you could use the archiving feature.  Above recommendations should be looked at for each camera.  Items like gain, sharpness, exposure (iris/shutter), compression, resolution, wdr, noise reduction, etc. can all be used to increase efficiency.  Look at the frame size (upper right of camera page) to see final bandwidth (look at this day, night, no/much activity).  Set cameras to VBR.

You can set up archiving on cameras you commonly need recordings from.  If you have motion or event setup properly (mask items that can false motion or use door contacts, PIR motion, etc. for event recording), you can have Exacq scheduled (after work hours) to archive event video to your site.  Use the client to access/search from the archive full speed.

 

CR
Chad Rohde
Jan 09, 2018

With an 8 to 1 ratio, depending on where this remote location is, a truck roll may be faster.

NH
Newell Hirata
Jan 11, 2018

Thank you all for the helpful comments and suggestions. Pulling the video back to our corporate office where I am needs to be at full resolution. 90+% of our videos are given to law enforcement as evidence of a theft so the video needs to be at the highest resolution possible to ID the suspect(s). This is my issue. As I stated in my original post 1 minute of video requires 8 minutes of real time. So if I need to pull 5 minutes as evidence of a theft it takes almost 40 minutes. Wanted to know if there may be other solutions other than increasing bandwidth between each location and my office.

Thanks Again.

 

JH
John Honovich
Jan 11, 2018
IPVM

Newell, thanks for the feedback.

so the video needs to be at the highest resolution possible to ID the suspect(s). 

Good point.

I want to emphasize there are definitely ways to keep the visible resolution the same while reducing bandwidth. For example, the fact that all video surveillance is compressed underscores that truth. The only question is how much compression and when / where to apply. Related: Video Quality / Compression Tutorial and How to Measure Video Quality / Compression Levels.

You do have the potential to keep the same visible resolution / evidentiary quality while reducing your bandwidth, i.e., my points above that the compression level could be set too low, requiring more bandwidth without visible improvements or that smart codecs could dynamically adjust compression level, reducing bandwidth without meaningful loss of details.

I am with you about being cautious but I do suggest you look a little more about these options in the future, as, e.g., smart codecs are a technology trend (much like years ago only highly bandwidth ineffecient MJPEG was used and then it improved to H.264, etc.).

 

 

(1)
UI
Undisclosed Integrator #3
Jan 11, 2018

I worked with a group that had very limited bandwidth between sites and had to produce an hour or more from each camera and surrounding cameras for an incident.

There were two solutions. 

First, they recorded both a high resolution stream and a very low resolution stream at the site. 

They could download the low resolution for the long time requirements remotely and a short period at high resolution for identification.

Additionally, each site had a client with an external usb drive they could Remote Desktop into and record it all in high resolution locally and ship or pickup the drive. 

This worked for them and their legal team. 

(1)
(1)
Avatar
Brian Karas
Jan 11, 2018
IPVM

Additionally, each site had a client with an external usb drive they could Remote Desktop into and record it all in high resolution locally and ship or pickup the drive.

Never underestimate the bandwidth of a station wagon full of tapes hurtling down the highway.

That quote is ~40 years old, but funny how even today it is still relevant that sometimes a wide area sneakernet is still the most practical approach. 

(1)
(1)
New discussion

Ask questions and get answers to your physical security questions from IPVM team members and fellow subscribers.

Newest discussions