HNHacker News
TopNewBestAskShowJobs

pktgen

820 karma · joined July 26, 2013

submissionscomments
pktgen··on Critical crypto bug leaves Linux, hundreds of apps open to eavesdropping
> And in other words, nobody should be writing anything in C. Ever.

And in other words, nobody should be listening to thrownaway2424. Ever.

pktgen··on How Telcos Will Make More Money, Cost Less To You, and Solve World Problems
Nah. Telecom lowering their price? LOL.

Normal businesses when their costs go down: since we're in a competitive market, we'll use this as a competitive advantage and lower our prices to match

Telecoms in the same: "we're sorry, but we've lowered our costs by $X. Unfortunately now we have the expense of handling all this new cash, so I regret to inform you we have to increase your bill by the same $X."

pktgen··on Why is Atom Closed Source?
The whole point is they're trying to have it both ways. They want to benefit from other people's work freely (that's why they're accepting pull requests), but do not want to let others benefit from their work in the same way.
pktgen··on Why is Atom Closed Source?
Love the double dipping. "GitHub can profit off your code, but you can't do anything with it and have no rights to it!"
pktgen··on Apple releases OS X Mavericks 10.9.2 with SSL fix
> Comcast seems unlikely to try to steal your bank password

I don't know about that. Comcast doesn't seem to have any problem robbing its customers right now.

pktgen··on Has goto fail been fixed yet?
Yeah, this is what I was thinking. elementaryOS seems to get the closest but still isn't ideal. I have to think about it. Thanks.

I use Linux on servers, but I've always found the options lacking in some way for desktop.

pktgen··on Has goto fail been fixed yet?
Just curious, what Linux distro is everyone switching to? At this point I am seriously considering it, because this is pathetic. (I suspect I'll remain with Apple hardware for the foreseeable future, because they still have the best laptops IMO, but running another OS is not out of the question.)

I like elementaryOS, but it really doesn't feel as polished as OS X. Things like their choice of font don't help IMO.

pktgen··on Comcast’s deal with Netflix makes network neutrality obsolete
Ideally that last mile network would be point-to-point FTTP though, eliminating this entire issue.
pktgen··on Anyone Who Still Thinks IPv6 Won't Happen Isn't Watching the Measurements
Expect this number to rise in the near future as Comcast Netflix traffic moves to IPv6: http://www.dslreports.com/forum/r29057080-
pktgen··on One of the Most Alarming Internet Proposals I've Seen
> Really? That's a special kind of evil and should be illegal.

Nah, no need to make it illegal. He's referring to AT&T's "GigaPower" gigabit service in Austin... the same service that's available from 2 other providers for the same, or cheaper, as what AT&T's charging for their data-mined bullshit.

AT&T will have a fun time getting customers when Google and Grande point that out in their attack ads. :D

pktgen··on FreeBSD 10.0 on Ubiquiti EdgeRouter Lite
Definitely. I am not sure what metrics Solarflare beat others (presumably including Intel) on, besides software support (OpenOnload implements the BSD sockets API and doesn't require software changes, while DPDK and others implement their own). In my testing, the X520s have been capable of both sending and receiving at line-rate.

It's possible the Solarflare NICs could be better at some metric like latency, by a matter of microseconds (only speculating here), but I can't see how they'd consider that worth the extra money for HTTP servers.

pktgen··on FreeBSD 10.0 on Ubiquiti EdgeRouter Lite
They do. They use Solarflare's OpenOnload: http://blog.cloudflare.com/a-tour-inside-cloudflares-latest-...
pktgen··on What to do after discovering SQL Injection vulnerability in random websites?
I would be very, very, very careful here. Not sure what country you're in, but you're setting yourself up for possible legal action, even though your intentions are good.
pktgen··on FreeBSD 10.0 on Ubiquiti EdgeRouter Lite
> 1. Data Plane Development Kit, which lets you skip the kernel IP stack (which takes thousands of CPU cycles to process) and do packet processing in userland taking just tens to hundreds of cycles per packet. http://dpdk.org/

I wish OS developers saw this as a problem. There is no reason kernel stacks should be so slow for tasks where all processing is done in the kernel. (For packets destined to userspace, you've got the syscall overhead to deal with.)

I recently tested the Linux network stack's PPS performance with an Intel X520 10GbE NIC. I used Debian testing, with the 3.12 kernel. My destination machine was an i7-3930K at stock speed. I wrote a simple kernel module adding an NF_IP_PRE_ROUTING hook returning NF_DROP with no processing, which would be the simplest possible code path. For a packet generator, I used another older machine with another X520, using the "pfsend" tool included with PF_RING, and the card in PF_RING DNA mode. That was easily able to saturate the link at line rate (14.8M PPS).

The result: the kernel was only able to sustain about 2.8M PPS.

I then loaded the DNA driver on the destination machine, used the included "pfcount" tool, and no packet drops - it was receiving the full 14M PPS.

I tested DPDK recently and had similar results.

I also modified the Linux ixgbe driver ixgbe_clean_rx_irq() function, and added a step in between the "fetch packets from RX ring" and "put packet in SKB and send to network stack" functions. Even when I added a bunch of useless comparisons for each packet, I was able to get ~12-13M PPS. I could get line rate by just dropping and not doing any processing.

pktgen··on FreeBSD 10.0 on Ubiquiti EdgeRouter Lite
UBNT is correct, but only for the hardware fast path. The driver for that appears to be proprietary.
pktgen··on FreeBSD 10.0 on Ubiquiti EdgeRouter Lite
UBNT's numbers are using the hardware fast path, which presumably is not supported by this project.
pktgen··on The New Normal: 200-400 Gbps DDoS Attacks
> Look at it this way. Without Cloudflare providing free DDOS mitigation, the attacks would not be happening as frequently.

I don't see the cost of CloudFlare's service as a problem. In fact, free DDoS mitigation is great. The problem is that they're willing to provide service to booters/DDoS-for-hire services.

Reputable providers prohibit this kind of activity in their AUPs, and there aren't that many companies with their own large networks capable of DDoS mitigation (many are just resellers). If CloudFlare stopped supporting such services, it would immediately put a large dent in the entire booter market because the kids would take care of shutting each other down, and individual booters would find it difficult to find a provider who can mitigate attacks and is willing to provide service to them. (In other words, let them fend for themselves. Screw them.)

> Maybe you'd feel differently if you were the target of some of these attacks.

I agree, and I think this is why a few of us (you and me, in particular) have different views compared to others.

(I have talked to you on WHT recently. Not under this name, but you might have an idea.)

pktgen··on The New Normal: 200-400 Gbps DDoS Attacks
I'm not and have never been a CloudFlare customer. My experience with them stems from hosting game servers and dealing with many DDoS incidents, nearly all of which originated from CloudFlare-"supported" (I would like to use the term "hosting" but I realize they will dispute this, and I'm not interested in a debate on semantics) booters. As part of this, I also have experience dealing with CloudFlare, which I detailed in another comment here.

Publishing research papers about DDoS attacks is one thing. Selling a service that performs them (i.e. DDoS-for-hire) is completely different, IMO.

pktgen··on The New Normal: 200-400 Gbps DDoS Attacks
Sure, you can _send_ spoofed packets from any host, but any reputable host will drop them.

Reputable hosts use uRPF or at least an ACL at their edge to drop any outbound traffic with a source address that isn't in their network.

People buy servers from Ecatel because they're one of the few that (intentionally) do not have such measures in place.

pktgen··on The New Normal: 200-400 Gbps DDoS Attacks
Firesheep is a poor comparison. It's a piece of software. The developer (or his associates) do not provide a product or service to protect against it, which would be a requirement to be considered racketeering. (Even if they were offering some "service," say to set up SSL on your server, I'd have a hard time calling it a racket simply because what Firesheep accomplished, i.e. packet sniffing for cookies, has always been possible.)

I don't have a problem with CloudFlare offering multiple service levels. That's just smart business. They are one of the relatively few companies with their own network that can mitigate attacks. And that's where the problem lies - they are using that capability to prop up booters (DDoS attack services, if you weren't familiar with the term). As others have said, booters would be largely uneconomical to operate without the cheap assistance of CloudFlare, because the booter "market" is similar to drug gangs - it's a "war" with the booter owners all trying to take out their competition. (So this makes DDoS attacks less easily available, a good thing for the rest of the Internet.)

The problem is that CloudFlare is choosing to allow them to operate, while offering protection services at the same time. This is the very definition of a racket. It is no different from "wouldn't it be a shame if your shop burned down, you should pay us money."

Now, I'm not saying that _all_ DDoS would instantly go away if CloudFlare stopped this. It wouldn't. It would make a significant difference though, IMO. Services like this make it easy for anyone with little skill to launch attacks, and with amplification techniques (NTP, DNS, SNMP, chargen, etc.) the booter needs very little hardware and bandwidth to launch massive attacks.

pktgen··on The New Normal: 200-400 Gbps DDoS Attacks
For something that threatens the very infrastructure CloudFlare operates their business on, I disagree.

Also, CloudFlare is not an essential service nor are they a monopoly. So if CloudFlare goes too far with something, it's not a problem for their customers to leave them.

pktgen··on The New Normal: 200-400 Gbps DDoS Attacks
Except CloudFlare only provides "basic DDoS mitigation" on their plans under $200/mo.
pktgen··on The New Normal: 200-400 Gbps DDoS Attacks
And here's what CloudFlare isn't telling you: many DDoS attacks are made possible thanks to their service, openly providing service (they would dispute my use of the word "hosting") to booters.

This is called racketeering. Create the problem indirectly and offer protection. Sound familiar?

pktgen··on The New Normal: 200-400 Gbps DDoS Attacks
I agree with you.

Having sent multiple abuse complaints to CloudFlare regarding booters, I have found them difficult to work with. As we've established, they will not censor anything; instead, if they determine your complaint to have some level of validity, they will send you to the actual host.

In one instance, the booter site had no information on registration or what was offered, so I gave them the hackforums thread where the service is being sold. I realize this is basically hearsay and not sufficient evidence to actually shut a site down, but remember that they won't shut a site down in any case. They did not consider this acceptable enough to release information about the host. They wanted me to register an account there and provide it to them for verification. Luckily, I could register without actually paying anything, providing me a nice UI with a big "launch attack" button, and this was sufficient for them.

More recently, they will not even release the site's IP address. All they will do is tell you to email the abuse department of [host] and ask the abuse department to contact them for details. This is ridiculous.

CloudFlare purports to be against DDoS attacks, yet has no problem providing service to admitted DDoS attack services. In other words, CloudFlare is a racketeering operation. They create the problem, indirectly, and offer services to solve it. (I realize they offer a free tier, but their advanced mitigation features are only available on paid tiers.)

pktgen··on The New Normal: 200-400 Gbps DDoS Attacks
Hosts that negligently allow (do not implement technical measures to block) packets to be sent from an IP address not routed to the sender.

Ecatel is the big one here. I don't know what it will take for their upstreams to shut them down, but it needs to happen. Do that and many of these reflected attacks will stop.

pktgen··on The New Normal: 200-400 Gbps DDoS Attacks
This is easier than you think. Most of these kids have no clue what they're actually doing and link their personal identities with the name they use.

I run (as a hobby) some game servers and have handled several DDoS incidents. Out of all of them, I think there's only been one that I couldn't find real-life details for. When this occurs and the perpetrator is a minor, and I can find details for them, I give the parents a friendly call and let them know. Usually this can be easily resolved in a peaceful manner and without any mention, threats or use of legal action.

pktgen··on NC Electric Car Drivers Begin Paying $100 Annual Fee
Why do taxes have to be spread out everywhere?

Why not just have an income tax and sales tax, or maybe only one (I've heard arguments for both sides, but don't want to get into that debate) to cover all this?

It's the same thing with all the ridiculous fees governments charge for various things. For example, registering a vehicle at the DMV. Why does this cost anything? After all, your taxes already pay the salary of the person at the DMV, their office, etc. and the sales tax (in states that have it) paid on a vehicle purchase should take into account that the new owner will need to register it.

It's just a ridiculous way of "not really increasing" taxes.

pktgen··on DoS attacks that took down big game sites abused Web’s time-synch protocol
Obviously such rules wouldn't automatically be in place; they'd have to explicitly be requested. And yes, it is a reasonable request to ask your upstream to add a few ACL lines.
pktgen··on DoS attacks that took down big game sites abused Web’s time-synch protocol
Correct. nLayer used to but stopped because Juniper's implementation is buggy.
pktgen··on DoS attacks that took down big game sites abused Web’s time-synch protocol
That's how Black Lotus does business. It's like the time when they were caught spamming and were made to look like complete fools on WHT.
← PreviousPage 7 of 8Next →