Subscriber Discussion

Pivot3 Edge Protect "Game Changer" Surveillance Appliance

Avatar
Brian Karas
Apr 04, 2016
IPVM
Pivot3 announced their "Edge Protect" appliance in March 2016. Among other things, Edge Protect was described as "a game changer", and claimed to provide "enterprise-class infrastructure without enter...

Read the full report here

Avatar
Jon Dillabaugh
Apr 05, 2016
Pro Focus LLC

Quad Gb NIC's, yet only 100Mb of processing power... brilliant!

(3)
UI
Undisclosed Integrator #1
Apr 05, 2016

The game appears unchanged part made me laugh. It's insanely expensive.

(2)
Avatar
Sean Nelson
Apr 05, 2016
Nelly's Security

I honestly dont see how they would sell even one of these at this price. The markup on these has to be insane.

Avatar
Brian Karas
Apr 05, 2016
IPVM

To be fair, they're not pushing these in a mainstream market, it's typically going to be companies that see the value in the whole "hyperconvergence" thing.

A friend of mine bought a new Lamborghini last year. One of the options added cupholders, and a couple of other interior bits and was $20,000. For most people $20,000 buys an entire car, in this case it was a "small charge" for some convenience items added to a larger product. He probably could have also gone to PepBoys and bought something that clips on a dashboard vent to hold his latte and saved $19,995.

I don't think Pivot3 will sell many (any?) of these as stand-alone servers, it doesn't make any sense. They might sell some that way to a customer who is planning to upgrade later but just wants to get something in quickly for now.

The more likely scenario is that a customer who has already bought into Pivot3's larger concept and has bigger servers deployed (the Lamborghini of this analogy) might buy these "cupholders" for some remote sites in order to keep their overall system consistent. Sure, they might be overkill compared to other options, but in the context of uniformity they can justify the price to themselves.

I would not anticipate many customers that aren't using Pivot3 hardware at other sites to look at this as a practical solution for their smaller/remote sites. Someone running a beefy Husky appliance at the main office would deploy a Husky appliance or 2 at the remote sites and call it a day.

In the past I was at another startup that sold "iron wrapped software" to a niche customer, similar to Pivot3's business model. It's tough because as a business you need to make some margin on the bare hardware, plus you need margin for your own software and for the build process (eg: technician time) to load it up. Even if you have an automated build process it can be more time consuming than you think. This naturally puts you into having to chase enterprise-class customers with bigger budgets. These customers have longer sales cycles, which increase the sales cost, which increases what you have to charge... In the end, you have to charge a LOT for what on the outside appears to be a run of the mill Dell/HP/Intel reference box with your "secret sauce" preloaded. Personally, I wouldn't want to chase that kind of business again.

(2)
(1)
U
Undisclosed #2
Apr 05, 2016
IPVMU Certified

Given the price of a stack of 3 Edge Protect units, a customer who wanted maximum resiliency could deploy a Windows-based Avigilon HD Appliance and a linux-based Exacq recorder in a dual-recorder architecture.

At the time of failure though wouldn't all the monitoring users go down until they switched to the other appliance and loaded a different vendors client program?

I'm assuming that the Pivot 3 cluster users would not notice nor have to do anything in event of failure. If it don't do that then it's pointless.

That said, I agree that most businesses would rather deal with a failover headache in case of the loss of a server; and $30,000 buys a lot of aspirin.

Avatar
Brian Karas
Apr 05, 2016
IPVM

At the time of failure though wouldn't all the monitoring users go down until they switched to the other appliance and loaded a different vendors client program?

No, we'd have fully redundant operators too ;)

Yes, my proposed solution gives cheap/simple resiliency at the cost of operator stress of switching over systems. It's highly unlikely anyone would really want that approach, but it does technically offer benefits that even the Pivot3 solution does not.

In reality, dual identical Husky (or Avigilon or Exacq or whatever) servers with some disk RAID, and a competent IT staff that monitors and maintains the machines, does backups/archives and so on would likely cover you against most realistic outages.

I like what Pivot3 is doing in concept. I've liked it since my friend first showed me a demo in 2003 of a similar concept at a stealth company he founded called Katana that later became Virtual Iron. The problem is that it's expensive. It's expensive to develop, it's expensive to setup the initial clusters, and it's expensive to have N+1 (or more) of your required compute resources to make the whole thing tick. Very few companies are able to justify the added expense for the incremental resiliency.

U
Undisclosed #2
Apr 05, 2016
IPVMU Certified

It's highly unlikely anyone would really want that approach...

No, I like it. It's the 'put each egg in a different basket' approach.

Much like when they select the tires on a newly commissioned jet how they intentionally choose from several different lots, to ameliorate the chances of a multiple tire failure on landing due to a single wonky production run.

Actually I don't know they do this, I just want to believe they do. ;)

Avatar
Brian Karas
Apr 05, 2016
IPVM

Pivot3's CMO provided the following response to this article in an email to us:

- There are significant benefits to a software-defined SAN that we believe are worth exploring in your analysis. All cameras have access to all storage regardless of which physical server they are associated with, and recorded video remains fully accessible even when an entire node fails. For customers for whom surveillance video is critical data, this is invaluable.

- We believe that the value of erasure coding is understated in the analysis as well. Using our patented techniques, we can protect data against the loss of up to 3 disks or an entire node plus one disk. This is not just RAID for servers, it is a much higher level of fault tolerance and a huge improvement over RAID.

- You have brought up throughput as an issue in the article. Our solution can scale up to 6 nodes/600 MBPS, and the traffic is automatically balanced across the cluster - this means that we scale very effectively and efficiently as end users' needs grow. So our system will handle roughly 600 Mbps of ingest, excluding live video redirect. One of the benefits is that it will sustain that throughput even during failures, a key differentiator from DAS. We also achieve this throughput at high fault tolerance levels (3 disk / 1 node + 1 disk) .... this is unique, since other DAS systems have to cut back their fault tolerance to achieve high throughput claims. We do not expect customers to have an issue with throughput with our systems based on our unique technology and approach; our benchmarks have borne out that most implementations of this scale in the market will be well served by our configuration.

We are also getting clarification on pricing and will update the post accordingly once verified.

(1)
U
Undisclosed #2
Apr 05, 2016
IPVMU Certified

So our system will handle roughly 600 Mbps of ingest, excluding live video redirect. One of the benefits is that it will sustain that throughput even during failures, a key differentiator from DAS.

So then assumedly, it will still do 300Mbps on a three node cluster with one failure.

And 2 standalone servers can each do 100Mbps X 2 = 200Mbps.

Yet 2 clustered servers do 300Mbps? How? Because of the shared drive array? There must be a better answer.

(1)
U
Undisclosed #2
Apr 07, 2016
IPVMU Certified

Forgive me but I'm confused about the pricing. In the UPDATE clarification you ask whether the $21000 is for one box or three. He seems to say "1". Yet you are saying it is for 3. Was the email recanted or am I reading it wrong?

New discussion

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

Newest discussions