Source Address Validation Everywhere, by Paul Vixie [APNIC 38 / Technical Keynotes]
•
0 likes•1,170 views
Source Address Validation Everywhere, by Paul Vixie. A presentation given at APNIC 38 during the Technical Keynotes session.
Recommended
More Related Content
What's hot
What's hot (20)
Viewers also liked
Viewers also liked (20)
Similar to Source Address Validation Everywhere, by Paul Vixie [APNIC 38 / Technical Keynotes]
Similar to Source Address Validation Everywhere, by Paul Vixie [APNIC 38 / Technical Keynotes] (20)
More from APNIC
More from APNIC (20)
Recently uploaded
Recently uploaded (16)
Source Address Validation Everywhere, by Paul Vixie [APNIC 38 / Technical Keynotes]
- 1. Source Address ValidationEverywhere Paul Vixie, CEO FarsightSecurity 2014-09-15
- 2. The Year 2002’s Biggest Problem? •“The most common attack on Internet hosts or infrastructure at the time of this writing is to cause the receipt of too much traffic, consuming all available resources on a victim's host or Internet connection. This is often called a "Denial of Service" (DoS) attack.” –ICANN SSAC SAC004, P. Vixie, October 2002 •(After RFC2267, P. Ferguson, D. Senie, January 1998) © 2014 Farsight Security, Inc.
- 3. Spoofed Source Attacks Internet target attacker reflector Srcaddr: (target) Dstaddr: (target) © 2014 Farsight Security, Inc.
- 4. Hopeless Trends •No incentive for up-front security engineering •No incentive for network output monitoring •Oft heard complaint: –“I’d be making all of the investment, but my competitors would be gettingall of the benefit.” •This is the “chemical polluter” business model –Externalized costs are downstream © 2014 Farsight Security, Inc.
- 5. Hopeful Sign: DNS RRL © 2014 Farsight Security, Inc.
- 6. Ode to David Isenberg •Rise of the Stupid Network, 1997: –“Why the Intelligent Network was once a good idea, but isn't anymore. One telephone company nerd's odd perspective on the changing value proposition” •David was right. We needed to innovate at the edge, and the core had to be assumption-free. •So, the core is stupid –like it has to be –But, so is the edge, which it must not be © 2014 Farsight Security, Inc.
- 7. So, Edge Device Quality? •Marketing & sales beats quality every time –Anybody can connect anything •QA budget shrinks at scale; only TTM matters –QA for a automobile tech: maybe $100/unit –QA for a Smart Phone: maybe $3/unit –QA for a CPE (cable/dsl/wireless): maybe $1/unit –QA for an embedded IoTdevice: maybe 5₵/unit •Note: 5₵/unit would be enough, iffup front © 2014 Farsight Security, Inc.
- 8. TCP Listeners as DDoSAmplifiers •TCP SYN occupies one octet of sequence space –TCP SYN+ACK, likewise –This requiredby TCP –TCP is requiredby the Internet •TCP requires retransmission until ACK –Including the SYN, and the SYN+ACK •So, every TCP listener is a 3x..20x amplifier –Problematic, even when not sent back-to-back © 2014 Farsight Security, Inc.
- 9. Technical Remediation •Near-end bandaids –Statefulrate limiting (e.g., DNS RRL) –Maybe TCP should only re-xmitwhen synch’ed? –Something’s got to be done about ICMP –…and about NTP and all other UDP protocols •Far-end solutions are far cheaper overall –Source Address Validation Everywhere (SAVE) –Make it the default; exceptions for multihoming © 2014 Farsight Security, Inc.
- 10. Nontechnical Remediation •Disrupt nation-state backed attackers –Some countries have earned Internet isolation •Increase compliance burden for device mfrs –Set a floor on quality and thus the QA budget •Increase compliance burden for ISP’s, telcos –Source Address Validation may have to be law •Consider Dan Geer’s recent proposal –A non-patchable embedded device would expire © 2014 Farsight Security, Inc.