BI 5 Server Setup for ~39 cameras + Deepstack

imjouster

n3wb
Joined
Jan 14, 2022
Messages
7
Reaction score
4
Location
Boise ID
Hey all, Been doing quite a bit of reading in the forums here trying to learn as much as I can about Blue Iris and just surveillance setups in general. I've got a friend that is building a new home on a pretty decent sized piece of property, and he has asked his brother and me to try to take on the project of setting up the property with surveillance and networking. This is going to be a big learning experience for both his brother and me. He has a background in networking, I've done my fair share of small networking jobs and building PCs for friends/family. But definitely a first for doing this size of a project. And a first for either of us using BI, so be patient if I'm not a BI Guru
:D
.

Main 2 questions I have... Is our hardware going to be capable of doing what we want. And do we go with WD Purple Pros that are running at around, $25-$26 per TB. Or should I go with Skyhawk AIs that I can get for around $23 per TB? The other option that just surfaced is getting Seagate Exos Enterprise drives, currently can get 16TB exos drives for around $19/TB. Also... has anybody had any experience in trying to mix WD Purples, Skyhawk, and/or enterprise drives? I know they have different softwares to make them ideal for surveillance drives, so I question mixing them.

Current planned setup (this may expand in future years and would like the system to be more or less future proofed)

35 4 MP anpviz PoE dome cameras. Most will probably run at 20 FPS with Deepstack AI being used. (have this figured to be ~38 GB/day usage per camera without Deepstack)
4 8 MP Starlight PoE Bullet Cameras. Running at 30 FPS with Deepstack AI. (have this figured to be ~90 GB/day usage per camera without Deepstack)

Server

4u Rosewill Chassis with 12 3.5" trays
Z590 motherboard
Intel i9 10850k CPU (10 core 3.6 Ghz)
32 GB (2x16gb) 3200mhz 16 Latency RAM (Actually have another 32 GB to throw at it to go up to 64 GB if needed)
500 GB m.2 Western Digital SN750 Black (for OS and BI only)
750W Platinum Seasonic PSU
NO Raid Array

Currently for storage we will either be going with WD Purple Pros, or Skyhawk AI drives, or maybe Seagate Exos enterprise drives. Going to start with ~150 TBs of storage and see how fast it fills up and go from there. I've tried calculating how much space is going to be required, and it's really tough to figure out without having some live feeds going and seeing how much is actually being used. If 150 TBs is overkill... that's fine... just means we can store stuff for longer (buddy originally wanted to store video for between 6 months and a year... talked him out of that...)

This is where things get kinda weird. My friend is going to end up having 3 main "zones" of cameras that will have different purposes.

There is a softball field and a baseball field that will be on the property that the "IDEA" is to be able to give parents access to those cameras to watch their kids play their baseball/softball games. Probably will plan on 1 camera per field at this time. There will also a large shop that will be broken into 3 areas (hitting facility for baseball, personal, and storage). Hitting facility will have the 8 MP cameras and once again we will want to allow parents to watch their kids practice through those cameras if they are at home and not able to attend the game or practice. We will call these parents "Group A". There will be 4 cameras in the personal area that we will allow certain people to log in to view. We will call these people "Group B". The rest of the cameras will be around the rest of the shop (inside and outside) and will just be used for general security cameras like the house.

Due to the fact that we are dealing with recording kids... we want to have at least 6 months of video storage on any camera that deals with kids so that we can protect ourselves from any future issues. The cameras around the house only need to store between 2-3 months of footage since they aren't dealing with kids.

So.... a summary of cameras and how long we will be storing footage...

15 4 MP (between 15-20 FPS) cameras around the main house. (2-3 months of footage stored) (no access from Group A or B)
4 8 MP (between 20-30 FPS) cameras in hitting facility. (6 months of footage stored) (Group A needs access)
6 4 MP (between 15-20 FPS) cameras for fields and general surveillance in hitting facility. (6 months of footage stored) (Group A needs access)
6 4 MP (between 15-20 FPS) cameras for personal side of shop. (2-3 months of footage stored) (Group B needs access)
8 4 MP (between 15-20 FPS) cameras for outside of shop and inside storage area. (2-3 months of footage stored) (no access from Group A or B)

Is the system that I have built going to be capable of doing all that we are asking of it? or should I consider building a 2nd identical system to help split the load? I know it would require a second BI license, but one reason we thought about doing 2 systems was just so we could have one system be dedicated to "the kids" and the 2nd one be for personal use.

Sorry about the long post... just really trying to make sure the whole situation was explained. Thanks for any advice and/or help.
 

bp2008

Staff member
Joined
Mar 10, 2014
Messages
12,666
Reaction score
14,005
Location
USA
I don't use Deepstack so I can't really speak for its performance, but I think you might benefit from adding a good GPU to offload the deepstack processing.

One system should be able to handle it all, and I think you have planned it excellently. But you will need to configure Blue Iris to use sub streams on most of the cameras (or all of them, ideally) so Blue Iris doesn't have to decode all those main streams all the time. You can also have BI send sub stream frames to deepstack and that will reduce the processing time required for deepstack too.

If you don't mind paying the premium for surveillance-optimized drives, feel free to get either the WD or Seagates, or even a combination of the two. But any properly functioning hard drive should be good enough and you shouldn't even notice a difference. I recommend you configure Blue Iris to record only a few cameras to each drive, that way the load on each one is relatively low and you don't need to worry about pooling the disks into one gigantic volume.

As for storage requirements, it is pretty straightforward to calculate. You are planning 35 cameras at 3.5 Mbps and 4 cameras at 8.3 Mbps. (35 * 3.5 Mbps) + (4 * 8.3 Mbps) = 155.7 Mbps. 155.7 Mbps * 6 months = 307 terabytes. That is, frankly, a ton of storage and your planned bit rates are not even all that high. So I don't think you're going to hit 6 months retention on a single 12-bay box unless you sacrifice bit rates (which I absolutely wouldn't recommend doing). One thing you might consider is "Continuous + Triggered" recording mode, where it will record the sub stream continuously, and record the main stream when motion is detected.

For something of this scale, I would recommend a second PC running a second copy of Blue Iris, but not to split the load. Rather to act as a backup. Computers in general aren't 100% stable, and Blue Iris is prone to crashing in some configurations. When it crashes when running as a service, it starts up again immediately, so less than a minute of video will be missed normally, but there is occasionally a complication. The crash can sometimes corrupt the clip database (thankfully rarely). It is possible to regenerate the clip database, but this prevents new recordings until the process is complete, and that can take quite a while. The one time I needed to do it, I had about 220,000 clips across one 2 TB and one 3 TB disk, and it took 27 minutes to regenerate the DB. It was actually of far more concern that, due to the database corruption, Blue Iris was deleting all my newest clips when the disks were full instead of deleting the oldest clips. It is what it is. Things like this are the reason I run a backup system.

So at my own home, I run two BI servers. The main box that I interact with daily records only on motion detection, so that increases how long my clips stick around by about 40x. It keeps video for so long that I just never bothered to upgrade its HDDs from the 2TB and 3 TB drives I have in it. Then I have a second BI server that records continuously but has only a fraction of the retention time. At one point I was getting about 10 days of retention despite having double the storage space of the box that was getting about a year of retention time. This backup box I do not interact with unless I want to check for something that happened recently that the main server missed. Because I don't use this box for live viewing or motion detection, I configured all the cameras with the "Limit decoding unless required" option and didn't bother setting up sub streams there, because substreams would just bring unnecessary complexity at that point.
 
Last edited:

Flintstone61

Known around here
Joined
Feb 4, 2020
Messages
6,587
Reaction score
10,894
Location
Minnesota USA
Just for an example,,,,as far as data storage frame of reference.....i have 18 cams split across an 8TB WD Purp 5600 RPM, and 5TB WD Blue 5400 RPM, with about 45 days of video to look back at. Most things around here that require a video review usually happen over the weekend and so I don't really need more than 30 days of data. Having said that...the George Floyd rioters procured the Post office keys to the City and were ripping off mail all over town everynite. Unbeknownced (spell) to us....they had been visiting our mail room and pilfering select letters to try and get checks and alter/copy them and turn them into cash.
This was one time where having 30 days of Video allowed me to show the Postmaster 2 of the incidents on the analog DVR cams. So I'd like to get 30-40 days if I can....I'm sitting on 2 more 8TB WD Purps, would like to add one to the system.....
 

Flintstone61

Known around here
Joined
Feb 4, 2020
Messages
6,587
Reaction score
10,894
Location
Minnesota USA
The Guru's in here often tried to tell me I don't need to run 20-30 FPS to capture stuff well. It took some practical experience with criminal or suspicious behavior to understand what they were driving at.
Now my LPR cams are running at 12FPS and a shutter speed of 1/2000. the lower FPS uses less storage, and doesn't compromise the quality of the images of the people and cars and delivery trucks in my 74 unit Condo application.
I was under the assumption that more FPS (30-60) was the answer to capturing sharp video. I'm learning a lot in here, thanks to the volunteers who offer good advice with new folks.
 

Flintstone61

Known around here
Joined
Feb 4, 2020
Messages
6,587
Reaction score
10,894
Location
Minnesota USA
as far as 2 Blue Iris, not a bad idea.....my home recorder is 9 cams on an Amcrest XVR, feeding identical data to my BI server, so i have redundancy. Not exactly what you propose, but just sharing some info with you.....
My work BI server records 18 of the 27 camera's, and the legacy DVR / analog recorder captures the older cams, and has, historically been more stable than the BI machine, BUT the caveat there is the user interface is Shite! and finding an incident is a Time killing pain in the ass. I can find incidents SO fast in BI now , I could never go back to the DVR/NVR UI as a daily driver......YMMV.....
 

SouthernYankee

IPCT Contributor
Joined
Feb 15, 2018
Messages
5,170
Reaction score
5,320
Location
Houston Tx
go with 15 FPS per camera.
Use BI substreams.
Get a good a nvidia card for deep stack.
If possible use a turret camera , bullet collect spiders, dome collect dirt and reflect light (IR)

Before going all in get one good quality variable focus camera and test each location, test at nigh with motion.
================================
My Standard allocation post.

1) Do not use time (limit clip age)to determine when BI video files are moved or deleted, only use space. Using time wastes disk space.
2) If New and stored are on the same disk drive do not used stored, set the stored size to zero, set the new folder to delete, not move. All it does is waste CPU time and increase the number of disk writes. You can leave the stored folder on the drive just do not use it.
3) Never allocate over 90% of the total disk drive to BI.
4) if using continuous recording on the BI camera settings, record tab, set the combine and cut video to 1 hour or 3 GB. Really big files are difficult to transfer.
5) it is recommend to NOT store video on an SSD (the C: drive).
6) Do not run the disk defragmenter on the video storage disk drives.
7) Do not run virus scanners on BI folders
8) an alternate way to allocate space on multiple drives is to assign different cameras to different drives, so there is no file movement between new and stored.
9) Never use an External USB drive for the NEW folder. Never use a network drive for the NEW folder.
10) for performance do not put more than about 10,000 files in a folder, the search and adding files will eat CPU and disk performance. Look at using a sub folder per camera (see &CAM in bi help)


Advanced storage:
If you are using a complete disk for large video file storage (BVR) continuous recording, I recommend formatting the disk, with a windows cluster size of 1024K (1 Megabyte). This is a increase from the 4K default. This will reduce the physical number of disk write, decrease the disk fragmentation, speed up access.
Hint:
On the Blue iris status (lighting bolt graph) clip storage tab, if there is any red on the bars you have a allocation problem. If there is no Green, you have no free space, this is bad.
======================================
 

imjouster

n3wb
Joined
Jan 14, 2022
Messages
7
Reaction score
4
Location
Boise ID
Alright Wow. Wasn't expecting this big of a response of help haha. Thanks to those of you who responded.

I don't use Deepstack so I can't really speak for its performance, but I think you might benefit from adding a good GPU to offload the deepstack processing.

One system should be able to handle it all, and I think you have planned it excellently. But you will need to configure Blue Iris to use sub streams on most of the cameras (or all of them, ideally) so Blue Iris doesn't have to decode all those main streams all the time. You can also have BI send sub stream frames to deepstack and that will reduce the processing time required for deepstack too.
Maybe what I'll do then is not worry about Deepstack at this time. I really know very little about it, I've just seen a lot of people using it. I'm sure it's something we could always add at a later date if we decide we need to. I'm really trying to avoid the thought of a dedicated GPU at this time due to the silly market we are in right now. And I do plan to use Substreams. And I'll definitely look into Trigger+continuous substreams. These are things I'm not sure I'm going to be able to know for sure what's best until we get everything setup and start recording.

You said something about using a 2nd server as basically a backup to the first. My only concern with that is I'm unsure of how you would set it up. You're going to have all the HDDs in one box connected through SATA to the motherboard. How would you get a second PC to connect to those same HDDs? I'm sure I could get the second server to find the cameras since they're all just going to be plugged into a PoE switch. Only way I can think of to do it off the top of my head would be to set up the HDDs as basically an NAS... but then wouldn't you have the problem of 2 blue iris's trying to write to the same drives and duplicating data?

The Guru's in here often tried to tell me I don't need to run 20-30 FPS to capture stuff well. It took some practical experience with criminal or suspicious behavior to understand what they were driving at.
Now my LPR cams are running at 12FPS and a shutter speed of 1/2000. the lower FPS uses less storage, and doesn't compromise the quality of the images of the people and cars and delivery trucks in my 74 unit Condo application.
I was under the assumption that more FPS (30-60) was the answer to capturing sharp video. I'm learning a lot in here, thanks to the volunteers who offer good advice with new folks.
That makes a lot of sense actually. I really thought 20 FPS would be the minimum I would want to go but for the general security cameras I think 15 or even 12 FPS as you just proposed may work. It's not really the smoothness of the video that you need. It's being able to have good quality clips when you need them. Are most cams able to vary the shutter speed? I know absolutely 0 about what shutter speed effects. I'd assume the faster the shutter speed the clearer the image?

go with 15 FPS per camera.
Use BI substreams.
Get a good a nvidia card for deep stack.
If possible use a turret camera , bullet collect spiders, dome collect dirt and reflect light (IR)

Before going all in get one good quality variable focus camera and test each location, test at nigh with motion.
================================
My Standard allocation post.

1) Do not use time (limit clip age)to determine when BI video files are moved or deleted, only use space. Using time wastes disk space.
2) If New and stored are on the same disk drive do not used stored, set the stored size to zero, set the new folder to delete, not move. All it does is waste CPU time and increase the number of disk writes. You can leave the stored folder on the drive just do not use it.
3) Never allocate over 90% of the total disk drive to BI.
4) if using continuous recording on the BI camera settings, record tab, set the combine and cut video to 1 hour or 3 GB. Really big files are difficult to transfer.
5) it is recommend to NOT store video on an SSD (the C: drive).
6) Do not run the disk defragmenter on the video storage disk drives.
7) Do not run virus scanners on BI folders
8) an alternate way to allocate space on multiple drives is to assign different cameras to different drives, so there is no file movement between new and stored.
9) Never use an External USB drive for the NEW folder. Never use a network drive for the NEW folder.
10) for performance do not put more than about 10,000 files in a folder, the search and adding files will eat CPU and disk performance. Look at using a sub folder per camera (see &CAM in bi help)


Advanced storage:
If you are using a complete disk for large video file storage (BVR) continuous recording, I recommend formatting the disk, with a windows cluster size of 1024K (1 Megabyte). This is a increase from the 4K default. This will reduce the physical number of disk write, decrease the disk fragmentation, speed up access.
Hint:
On the Blue iris status (lighting bolt graph) clip storage tab, if there is any red on the bars you have a allocation problem. If there is no Green, you have no free space, this is bad.
======================================
Like I said to bp2008, might just not worry about Deepstack at this time. I guess the biggest thing on that is what am I going to be missing out on if I don't run Deepstack? Am I making a big mistake if I decide not to use it?

I think I've seen you post that Standard allocation post in the past. And I'm pretty sure I even have it bookmarked as a guideline to setting up BI and getting it all optimized. I do plan to use Substreams. The cameras are going to be mostly Dome cameras. For better or worse... all the cameras are already purchased, so going to have to work with what I have on that for now.

I've kind of got things mapped out right now with the cameras to where I have 5 designated "zones" (for the general security anyway) that I plan to have 1 zone per hdd. If there are 2 cameras on the front porch, the are on different zones, so that if one HDD fails, I still have the other camera I can look at. Which is what you were suggesting in #8, so that's comforting to know.
 

wittaj

IPCT Contributor
Joined
Apr 28, 2019
Messages
24,428
Reaction score
47,550
Location
USA
Stay away from the consumer big box type cams and cheap no-name cams and yes you can set shutter speeds. Avoid Reolinks and any system that is proprietary like Ring, Nest, Arlo, etc.

Shutter is more important than FPS. It is the shutter speed that prevents motion blur, not FPS. 15 FPS is more than enough for surveillance cameras as we are not producing Hollywood movies. Match iframes to FPS. 15FPS is all that is usually needed.

These types of cameras are not GoPro or Hollywood type cameras that offer slow-mo capabilities and other features. They "offer" higher FPS to appease the general public that thinks that is what they need, but you will not find many of us here running more than 15 FPS; and movies are shot at 24 FPS, so anything above that is a waste of storage space for what these cameras are used for. If 24 FPS works for the big screen, I think 15 FPS is more than enough for phones and tablets LOL. Many of mine are running 10 and 12 FPS - goal is a clean still image capture, not butter smooth motion.

If you purchase cameras with AI, then you don't need Deepstack. Deepstack will give you a nice orange figure or vehicle to let you know it triggered on that, but with the AI in the cams, it is sufficient for most needs. Deepstack is much more critical for non-AI cameras. But even with AI cams, there are still use cases for using Deepstack with it. That may take it down from 39 cameras using Deepstack to 2 or 3 cameras.

 

imjouster

n3wb
Joined
Jan 14, 2022
Messages
7
Reaction score
4
Location
Boise ID
So here's where I'm at on HDDs right now. I can pick up the following HDDs for the following prices... and I'm really fighting what route to go... I'm trying to make sure I buy either directly through amazon prime (not fulfilled by someone else), Sold by Newegg, Direct from WD.... I've seem some really cheap prices for drives... and some of the sellers have terrible ratings... So trying to avoid them at all cost.

All these drives are 7200 RPM with 256 MB of cache and 550/TB year durability ratings.

Seagate Exos 16TB Enterprise
$305 which is $19.06/TB (direct from Newegg)

Seagate Skyhawk AI 16TB
$399.99 which is $25/TB (Amazon Prime)
$279.99 for 12TB which is $23.33/TB (BH Photo)

WD Purple Pro 12TB
$329 which is $27.41/TB (direct from WD)
$361.78 for the 14TB which is $25.84/TB (amazon prime)

So is there really any reason NOT to go with the Enterprise drives? $4/TB less than the skyhawk and $6/TB less than the purples... That's quite a chunk when we're talking about 150+ TBs of storage.
 

imjouster

n3wb
Joined
Jan 14, 2022
Messages
7
Reaction score
4
Location
Boise ID
Cameras are going to be Anpviz 5MP dome cameras for the most part (36 of the 39). IPC-D250W-S is the model number on them. We've already purchased all the cameras... so gonna have to work with what we've got fortunately/unfortunately. I can't tell for sure if these cameras are AI or not... they claim "motion detection" But I doubt that qualifies as AI. If they are not AI then how critical would you consider Deepstack to be?

Also the only cameras I think we are going to try to run 30 FPS on are the 8 MP bullet cameras, and the ones covering the fields. Those are going to be meant for being watched on other devices by parents, and probably will never need to take a "still" from those cameras. They are almost purely for entertainment purposes... which I'm still not sure how well that is going to work, but we shall see. I know the cameras aren't go pros or anything like that, but I'm hoping they'll work slightly better than our current solution which is an Iphone on a tripod streaming to facebook live...

Stay away from the consumer big box type cams and cheap no-name cams and yes you can set shutter speeds. Avoid Reolinks and any system that is proprietary like Ring, Nest, Arlo, etc.

Shutter is more important than FPS. It is the shutter speed that prevents motion blur, not FPS. 15 FPS is more than enough for surveillance cameras as we are not producing Hollywood movies. Match iframes to FPS. 15FPS is all that is usually needed.

These types of cameras are not GoPro or Hollywood type cameras that offer slow-mo capabilities and other features. They "offer" higher FPS to appease the general public that thinks that is what they need, but you will not find many of us here running more than 15 FPS; and movies are shot at 24 FPS, so anything above that is a waste of storage space for what these cameras are used for. If 24 FPS works for the big screen, I think 15 FPS is more than enough for phones and tablets LOL. Many of mine are running 10 and 12 FPS - goal is a clean still image capture, not butter smooth motion.

If you purchase cameras with AI, then you don't need Deepstack. Deepstack will give you a nice orange figure or vehicle to let you know it triggered on that, but with the AI in the cams, it is sufficient for most needs. Deepstack is much more critical for non-AI cameras. But even with AI cams, there are still use cases for using Deepstack with it. That may take it down from 39 cameras using Deepstack to 2 or 3 cameras.

 

wittaj

IPCT Contributor
Joined
Apr 28, 2019
Messages
24,428
Reaction score
47,550
Location
USA
Most here do not have good success with domes outdoors, but I guess you gotta go with what you go with. And yeah those cams do not have AI.

Keep in mind up until about a year ago, Deepstack was not integrated with Blue Iris. Up to that point, it came down to dialing in the motion setting for each camera and each field of view to knock out the false triggers. Most of us here spent the time to dial in each camera to eliminate the false triggers. I rarely had a false alert prior to Deepstack because I dialed in the motion settings.

Deepstack is simply another tool in the toolbox and you can absolutely run the system without Deepstack. Depends on what the end use is.

You may find the iphone on a tripod works better LOL.

If your unique case requires that type of FPS, you will find surveillance cameras are not going to meet your needs and you need to get a camera capable of that - or spend some serious money.

We have had recently people come here after purchasing cameras in two instances where they were wanting 60FPS - one was a tennis club and another was a youth soccer club. In both cases they found that these types of cameras were not capable of what they were wanting to do. Sure the cameras could run faster FPS, but it still didn't provide them with the level of detail they were looking for. I recall the soccer club had a decent quality PTZ ($800) that is fine for a residential or retail/commercial installation, but to cover the action of the soccer field it wasn't capable of meeting their needs. And because of the extremely fast motion, it was creating a halo type effect around the action (which can be seen in certain lighting conditions). These cameras are good, but not good enough to catch the rotation of a ball for example.
 

imjouster

n3wb
Joined
Jan 14, 2022
Messages
7
Reaction score
4
Location
Boise ID
Unfortunate to hear on the domes... guessing the biggest issue is just keeping them clean from dust. Might be on the monthly chore list for his kids :D. The camers were recommended to us by a guy who I guess does surveillance systems in new residential homes. He's been willing to give us recommendations. Just isn't going to be a part of putting it all together. Might be on the monthly chore list for his kids :D.

Yeah so the cameras for the parents to view, as long as they can see the main baseball diamond, and their kids running the bases... swinging the bat, etc. that's all we really need out of them. There will also be the 3 higher resolution cameras inside the hitting facility for parents to watch their kids. But it's nothing like trying to track ball spin or anything like that. The baseball clubs already all pitched in and bought Rhapsodo machines which I guess are pretty dang cool cameras. They will have nothing to do with my blue iris setup though. Our setup for for the parents to view is literally just for them to be able to kinda "be there" without actually being there. And if it doesn't work... I'll repurpose the cameras elsewhere hah!.
 

wittaj

IPCT Contributor
Joined
Apr 28, 2019
Messages
24,428
Reaction score
47,550
Location
USA
We have found here most of the professional "guys" installing these systems don't really know what they are talking about LOL. They install domes and Night Owl kits bought at Costco LOL.

Domes are not recommended unless it is well protected from the elements. The big issue is that the domes will hold onto water droplets much longer than a turret or bullet and thus any dirt on it can then become problematic. And during a rain event, the camera is basically blind...

1642397169863.png

Domes close to a baseball field will be trouble LOL. Lots of dust. Then rain. Then dirty rain spots. Glare from all angles of the sun. Then over time the sun will "fog" the dome like a car headlight. At least they are "cheap" cameras, but when buying 36 of anything it adds up quick!

Sounds like a good chore for the kids as you said LOL.
 

bp2008

Staff member
Joined
Mar 10, 2014
Messages
12,666
Reaction score
14,005
Location
USA
Maybe what I'll do then is not worry about Deepstack at this time. I really know very little about it, I've just seen a lot of people using it. I'm sure it's something we could always add at a later date if we decide we need to. I'm really trying to avoid the thought of a dedicated GPU at this time due to the silly market we are in right now. And I do plan to use Substreams. And I'll definitely look into Trigger+continuous substreams. These are things I'm not sure I'm going to be able to know for sure what's best until we get everything setup and start recording.
I don't blame you for not wanting to buy a GPU right now.

You said something about using a 2nd server as basically a backup to the first. My only concern with that is I'm unsure of how you would set it up. You're going to have all the HDDs in one box connected through SATA to the motherboard. How would you get a second PC to connect to those same HDDs? I'm sure I could get the second server to find the cameras since they're all just going to be plugged into a PoE switch. Only way I can think of to do it off the top of my head would be to set up the HDDs as basically an NAS... but then wouldn't you have the problem of 2 blue iris's trying to write to the same drives and duplicating data?
It wouldn't be much of a backup if it used the same storage devices. The idea behind a backup BI server is it is a completely separate PC with its own hard drives. That way, any failure of one PC does not affect the other.
 

sebastiantombs

Known around here
Joined
Dec 28, 2019
Messages
11,511
Reaction score
27,690
Location
New Jersey
A GPU suitable for use with DeepStack can still be bought for $200-$300 dollars. A Quadro or GTX10xx will work very well and can be had on the used market in that price range. With the overall cost of your system it's not a lot of money. I would invest in a UPS capable of running the system during short power failures and capable of shutting it down gracefully in extended power failures.
 

imjouster

n3wb
Joined
Jan 14, 2022
Messages
7
Reaction score
4
Location
Boise ID
I don't blame you for not wanting to buy a GPU right now.



It wouldn't be much of a backup if it used the same storage devices. The idea behind a backup BI server is it is a completely separate PC with its own hard drives. That way, any failure of one PC does not affect the other.
Sorry about the delay in responding here. been a crazy last week and a half....

Ok yeah, thought you were just wanting to backup the PC components minus the HDDs. The HDDs by themselves are going to end up costing well over $3000 for one system. And to have 2 systems with the same amount of storage just won't be in the cards unfortunately. Do definitely plan to have windows updates turned off. I thought you were just talking about failures like power supplies, ram, boot/database ssd. I know things fail in a PC unfortunately. Typically though the highest failure rate is HDDs in my experience. So hopefully the setup I have will help mitigate the effects of a single HDD going down. This is one of the reasons why I'm trying to "zone" the HDDs so that if one drive fails, I have 2-3 other HDDs with cameras looking in the same general areas so hopefully we can still

We will have a UPS of some sort in the rack. Most likely going to be the Tripp Lite SMART1500RM2U.

So realistically... and I know this is said by everybody before they put up a security system. I don't expect to have to pull footage on these cameras really ever... It won't be unless a parent complains "this person molested my kid" or we have a break in at the house, or something like that that we'll actually have to be pulling footage. So if the system goes down, and we lose footage for a week. 99.999% chance it won't matter.

With that all being said. what I would be able to convince my buddy to do is to build a 2nd server... but the storage would be split between the 2 servers. already was figuring if I had 2 cameras pointing in the same general area I'd have them stored on different HDDs. I suppose I could always have those 2 different HDDs on the different servers. This is all assuming we do the 2nd server which is kind of up in the air... we really only wanted to do the 2nd server if we needed to split the load of the cameras. We have a 10Gbps fibre cable ran between the house (which is where the main "server room" is) and the "server room" in the shop. So moving data between the 2 server rooms won't be an issue.
 

bp2008

Staff member
Joined
Mar 10, 2014
Messages
12,666
Reaction score
14,005
Location
USA
A backup server could just record on motion trigger and have a single hard drive.

Thats right, 99.999% of the video that is recorded is worthless. So it is really more of an academic exercise. Extra money and complexity just for a teeny tiny bump in reliability.
 
Top