Follow Slashdot blog updates by subscribing to our blog RSS feed


Forgot your password?
Check out the new SourceForge HTML5 internet speed test! No Flash necessary and runs on all devices. ×

Comment Re:Set up correct secondary DNS servers (Score 1) 321

Secondary DNS would not have helped here. The issue with DNS is that it's a centralizing service.

I understand that you have a particular drum to beat in this regard, but the problem is actually that Dyn hosted both the primaries and the secondaries, and they took Dyn offline.

If the primaries were at Dyn, and the secondaries were not at Dyn, none of the sites would have experienced any downtime.

Comment Re:Set up correct secondary DNS servers (Score 1) 321

So my IoT thing sends out a http request on port 80 of your web server, is that a DDOS attack or is that a valid request?

In my personal opinion?

It's always an attack, since IoT devices should connect to an Intranet server under your control, and not be vended routable addresses under any circumstances.

Comment Re: Set up correct secondary DNS servers (Score 1) 321

If your TTL is high enough, attacking a DNS service wouldn't deny service. The RFC says at least 1800s. Most of these sites have such poor uptime/architecture that their TTL is set to 120 or less.

Most caching servers at ISPs are set up in violation of the RFCs anyway:

* If they do not have an IPv6 upstream, they fail to filter IPv6 addresses out of their responses to downstream DNS requests.

* If they get some TTL value with less than their idea of a "minimum", they modify the TTL to be 300 or more seconds.

The first makes it hard to be "IPv6 by default", i.e. listing the IPv6 responses first in preference order over the IPv4, since it makes it not work for some people on the downstream side (the IPv6 addresses have to each time out before an IPv4 address, if there is one, is attempted).

The second makes it a real time consuming thing to do to have to wait 5 minutes between testing DNS reconfigurations to see if they work (and then you get 5 minutes of downtime when they don't, before you can fix them).

Comment Incorrect. 10,000 DNS servers in the pool... (Score 1) 321

But I will. If you spit it up into two sections, then the attacker will simply attack both servers. How many secondary servers would you need before the attack is spread too thin to deny service? Who knows.

That's easy. You put ALL of them in the peering pool. If you don't put your servers in the peering pool, then an attack can take you down... but no one else. Good luck getting customers in the future.

It's very easy: 10,000 DNS servers means a 1:10,000 chance of them hitting both your primary and secondary servers for your domain. Unless it's YOU the bad guys are attacking, instead of the DNS infrastructure (and if it's YOU, you have other problems), then it's unlikely that both your primary and secondary will get hit.

But don't forget that the companies are paying for all this bandwidth.

Yes. And to make it fair, you scale your presence in the pool by the number of domains you are personally hosting. If you host 1,000 domains, then at most you will also be secondary for 1,000 domains. If you host 1,000,000 domains, then you will host at most 1,000,000 secondaries.

This is why it's a peering pool.

Even if their services stay online they're spending $$$ to keep them online while the attacker isn't spending any money.

One company is an acceptable casualty. It's likely, however, that the Bad Guys(tm) were either targeting a number of specific domains, or they were targeting Dyn itself.

Either way, you'd set up collective defense resources for all pool members (that way, even if they were just going after Dyn, you could still afford to go after the culprit).

Comment Blame the ISPs (but especially Dyn) (Score 1) 174

Properly configured DNS secondaries hosted at different ISPs would have completely mitigated the problem for everyone but Dyn. Because Dyn hosts its own secondaries, hitting Dyn downed both primary and secondary servers.

ISPs need a peering pool arrangement for DNS secondaries, where secondaries are distributed over the entire pool.

This is how it was designed to work: multiply connected redundant secondaries.

The worst damage possible in that scenario is the inability to update DNS information hosted at Dyn itself, or to initiate zone transfers in or out of Dyn.

That reduces it from an attack on the DNS infrastructure to an attack on Dyn itself (which is much less important to everyone but Dyn).

Comment Set up correct secondary DNS servers (Score 5, Interesting) 321

Set up correct secondary DNS servers.

If the secondaries had not been hosted at the same company, but instead at various companies around the world, the attack would have had no effect on anything but traffic.

This is, by the way, how multiply connected networks are supposed to work.

This could be easily accomplished at no additional cost by having a peering-pool arrangement between all the host registrars, so that we ended up with a multiply connected redundant network.

Kind of how we designed the thing to work in the 1960's and 1970's, and DNS itself in the 1980's.

But a lot harder for law enforcement to issue DNS-based takedowns on, of course. Since it would route around the damage and keep functioning. As designed.

Comment Re:It was a premises warrant. (Score 1) 420

He may in some cases be forced to surrender a key to a strongbox containing incriminating documents, but I do not believe he can be compelled to reveal the combination to his wall safe —- by word or deed

Meaning the fingerprint gathering for the use of opening the phone is tantamount to compelled testimony in the general case, while the fingerprint gathering for the use of identification and matching is not.

Keys don't change. Fingerprints don't change. A biometric identifier is therefore not affirmative.

Combinations can change. Pin codes can change. Utilizing either requires active participation in a process. And is therefore affirmative.

Fingerprint usage is therefore tantamount to using a key, and if you are stupid enough to use a biometric identifier as an access method, you've picked a non-affirmative access method.

Comment It was a premises warrant. (Score 5, Informative) 420

In a premises search, they can compel an unlock of phones by fingerprint, assuming you lock your phone that way.

The specific legal decision was the 1988 John DOE, Petitioner v. UNITED STATES. 487 U.S. 201 (108 S.Ct. 2341, 101 L.Ed.2d 184) decision.

It came down to whether on not an affirmative action was required on the part of someone, or if it was a non-affirmative action. Use of a key on a safe or lockbox is not affirmative. Being forced to enter the combination is not affirmative; it's tantamount to compelled testimony.

Here's the part of the decision of interest:

A defendant can be compelled to produce material evidence that is incriminating. Fingerprints, blood samples, voice exemplars, handwriting specimens, or other items of physical evidence may be extracted from a defendant against his will. But can he be compelled to use his mind to assist the prosecution in convicting him of a crime? I think not. He may in some cases be forced to surrender a key to a strongbox containing incriminating documents, but I do not believe he can be compelled to reveal the combination to his wall safe —- by word or deed.

Moral of this story: use a pin code, rather than using the fingerprint unlock. It may be a cool feature, but it offers you no legal protection.

Comment "Wouldn't they be stuck in the traffic as well?" (Score 0) 60

"Wouldn't they be stuck in the traffic as well?"

I believe the theory is that if you practice something, you get better at it. An Uber driver (presumably) practices driving, which means they get better at it, which means that they don't automatically slow down any time they see a huge ball of fire in the sky (try 101 Northbound at 4-5 PM), or other stupid things that less practiced drivers do, meaning they end up not clogging things up, like less practiced drivers tend to do.

The expression "Sunday driver" is actually based on observations.

Comment They already know the cause. (Score 4, Interesting) 145

The fact that they can't determine why these phones are going up in smoke is scary. In a way it's understandable; the ones that do end up exploding burn up so there's no system logs or other evidence that could be checked to determine the cause.

The problem is obviously the charging circuit. If it were anything else, they could just put in better batteries, or ship better chargers. The recall happened because the problem is on board the phone itself.

Newer phones still have the problem, so we know it's a design problem, rather than a component sourcing problem (like the counterfeit capacitors problem). In addition, Samsung manufactures their own phones, and their assembly lines operate differently, compared to Chinese assembly lines at Foxconn: it's very easy for them to localize a problem in the manufacturing process, whereas Foxconn goes out of their way to hide it by making bad employees into nameless cogs.

So basically, they have a design problem in the charging circuit, probably in the cell leveling portion of the charger, in the same way that the "Hoverboard" clones that keep starting on fire have a known bad charging circuit that overcharges some lithium cells in the larger battery, while other lithium cells get too little charge, on the charging circuit keeps drawing amps for all of the cells.

Then when the overcharged cells are discharged, they pretty much "Flame On!", and someone does a fair imitation of The Human Torch(tm).

This stuff isn't rocket science, it's basically third year in a U.S. community college EE and analog circuit design.

Comment Re:really (Score 1) 145

Yes, there's an issue with software updates .... vendors need to be more responsible about that.

Vendors are. Cell carriers aren't. Without a new shiny, how are they supposed to lock you into a new contract for another 18 months, until the next new shiny comes out?

Everything is predicated on locking a customer into your business, in order to reduce customer acquisition costs. It costs a heck of a lot more to acquire a customer than it does to lock them into a contract so you can retain them.

Without this aspect of the business model, both your cell phone costs and your service costs go up. The phone costs go up because they are no longer subsidized, and the service costs go up because they can't amortize the customer acquisition across an average of 5-7 years, and instead have to worry about the customer leaving.

The entire telephone company service model has always been about charging based on circuit switching points, and charging for long distance. Now that everyone is using cell phones, they can't do that any more, and have moved to packet switched networks. But in order to maintain their profit margin, they've had to push the costs off to other areas.

In case you care, most of the costs come from federally mandated rural service. If the telephone companies didn't have to provide service so that when someone in a rural area was having a heart attack, they didn't just conveniently die, and not be in a rural area any more, they could vastly reduce their infrastructure costs.

Most of the rest of the world (certainly Europe) doesn't seem to realize that the U.S. has about 180,000,000 people who do *not* live within 50 miles of a coastline. Unlike the U.K., where *everyone* lives within 50 miles of a coastline. The U.S. is *big*.

Slashdot Top Deals

Why won't sharks eat lawyers? Professional courtesy.