Hacker News new | past | comments | ask | show | jobs | submit login
Ex-CIA Agent Raises $40M to Find Every 'Thing' on the Web in Just One Hour (forbes.com/sites/thomasbrewster)
82 points by rbanffy on Aug 31, 2017 | hide | past | favorite | 38 comments



Shodan (https://www.shodan.io/) been doing this for over 5 years, with a similar tagline: Shodan is the world's first search engine for Internet-connected devices.


From the article: "It's comparable to Shodan, a search tool for connected devices, but turbo charged and closed to the general public, Junio's previously said."

I think their big thing is speed. They are scanning the entire IPV4 space in 60 minutes which... is kind of impressive? I haven't thought about it but doing a port scan on my public network takes a little bit of time and I'm only dealing with couple dozen IPs.


ZMap claims to scan the IPv4 space in 45 minutes (although you would need some fancy hardware for that ;)


> although you would need some fancy hardware for that

You only need 1 Gbit/s upstream and a laptop to hit the 45 minute mark. On fancy hardware zmap needs less than 5 minutes.

One of the people behind the project presented it a 30C3 a few years back (including a live scan) in case anyone is interested: https://media.ccc.de/v/30C3_-_5533_-_en_-_saal_2_-_201312281...


Oh neat, I thought it was a 10G card they were using. Wow, 5 mins is amazing with 10G.

I'm just browsing their website some more, they seem to have lots of interesting libraries too, like https://github.com/zmap/ztag

"ZTag processes ZGrab output and annotates raw scan data with additional metadata such as device model and vulnerabilities."


I wonder how many years it will be until we can have a personal search engine, like google , running on our own server.

Surely the technology will be cheap enough to do at some point.

(To reach to the capability of what google is doing right now, not to surpass what they will be doing then )


Thanks for reminding me about the Ztools! Absolutly love some of the applications found on https://zmap.io/.



"Right now, Qadium can reach every connected device in the IPv4 space -- made up of the the millions on millions of IP addresses of web devices"

So can my phone?


They literally just described the internet.


> That massive-scale scanning (or what Junio prefers to call web-scale sensing)

At what point does tech become a parody of itself?


I prefer to call it machine-vision infused cyber-fabric deterministic detection


Thanks, guess they'll add that to the homepage in 60 minutes


Skills required: MVICFDD (10 years)


I think the key quote is "The ultimate aim of the service is to help companies determine if there are vulnerable devices on their network that could be exploited by malicious hackers, who could then pivot and compromise the whole organization."

I believe the founder was on Recode podcast within the last year. It was a great listen and really sold the product value prop. I work in healthcare and in my office alone we have 2 or 3 printers connected that IT isn't aware of because we couldn't get approval.


And with their speed they could probably run it every hour so that IT can act immediately on potential vulnerabilities, not just testing it every few months.


For those who it matters:

He wasn't a case officer (AFAICT), he was not an agent (somebody an officer recruits to spy), he was an analyst.

The distinction is important.


What's the advantage over Shodan out of interest, or Zmap?


This sounds like nothing new, other than a new way to take money from ignorant executives.


Yes, but it is new to ignorant executives. There is always a business selling an old thing with extra marketing.


I learned about Qadium last spring. Not sure how under the radar they were then, but I heard about it through a referral of a referral working on cyber risk.

If they are able to credibly provide a baseline for number of connected devices and percentage that are vulnerable, it won't be a huge step up from zmap, but it could provide some interesting data for security pros, regulators and insurance. Not sure how they get to the part about vulnerability of devices.

The cyber kill chain provides a nice jumping off point for a hierarchical model, but the kind of data necessary to really flesh it out is just so damn hard to find at the quality levels necessary.


It's nice Peter Thiel found a new hobby other than testing hepatitis treatments in low income countries that is illegal in the US. Even though this is a scam and as many other people pointed out, something anyone can do, it's actually one of the less shady things Peter Thiel has done. Good job Peter. Way to be somewhat less shady.


Am I correct in understanding that this just pops off a list of IP addresses reachable from whatever "point" you start the "scan" from?

Is there more info into "how"? How is this something that doesn't already exist? Other's have referenced Zmap, how is this an improvement over it, or even just a patient person with Nmap?


"Junio is particularly proud that none of his customers were infected with WannaCry, the ransomware"

Uhmm.... so, it detects WindowsXP installations?


The majority of infected computers form WannaCry was windows 7, its a common misconception that windows xp was the most hit. But never the less I don't see how this scanner could of prevented that unless he added EnternalBlue etc to the scanner once smb was found to be open.


Am I correct in assuming that this becomes a total waste of $40m the moment people start finally transitioning to IPv6?


Eh, not really, it's just going to require an exponential increase in computing power. They already have the platform, in theory it would just be about scaling.

That said, it's kind of a huge scaling problem. If it really is a matter of throwing more hardware on it, and they currently use, say, 100 servers, they're going to need 429496729600 servers to get the same times they are getting now.


IPv6 is 2^96 times larger than IPv4. You'd need 100*2^96 servers. That's not tractable with a brute force method.


Maybe not that many, but still a lot more, if you first looked at BGP, and only brute-forced networks that were routable, rather than the entire search space.

Basically, before scanning, query and iterate all ASes advertisments and union present ranges, then feed that to zmap or something that can scan IPv4 and/or IPv6.


I dunno how I messed that up. I'mma blame the post-lunch slump. For some reason I was thinking IPv6 was 64 bits.


Does it not get worse the larger it gets or no?


If you google the number of Possible IP (v4) addresses you get an answer that kind of gives you more context for your question:

IPv4 uses 32-bit IP address, and with 32 bits the maximum number of IP addresses is 232—or 4,294,967,296. This provides a little more than four billion IPv4 addresses (in theory). The number of IPv4 available addresses is actually less than the theoretical maximum number.

So you would need to be prepped to hit 4 billion approx devices.


It would be good to know how many companies are using reputable addresses internal, unreachable from the open internet based on deduction


Yeah guess would need more context I mean I hear crazy processing times l (requests per second) whether it's go or visa transactions it seems possible with enough resources but what do you do/compare I don't know haha. IPV6 too.


So nmap? Okay...


https://zmap.io/ is the tool of choice for scanning IPv4 space quickly. If you have good connectivity and the correct hardware you can scan a lot of space very very quickly.


$40m nmap


So, build Google's index from scratch under an hour.

Good luck!




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: