1 / 23

I don't like spam!

Spam, spam, Steve VanDevender, bacon, and spam. I don't like spam!. Spam defined, spam, spam and spam. Our working definition of spam: Unsolicited Bulk Email Unsolicited Recipients did not request mail from sender Bulk Single message sent to many recipients

quito
Download Presentation

I don't like spam!

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Spam, spam, Steve VanDevender, bacon, and spam I don't like spam!

  2. Spam defined, spam, spam and spam • Our working definition of spam: Unsolicited Bulk Email • Unsolicited • Recipients did not request mail from sender • Bulk • Single message sent to many recipients • Spam is about consent, not content • Spam is not just commercial or advertising email, but any unwanted bulk delivery

  3. Spam vs. legitimate bulk email • Solicited bulk emailing is generally accepted • Recipients request subscription directly • Subscription is confirmed so parties cannot add unwilling recipients • Subscriber must reply to confirmation message sent to subscription address to complete subscription • Subscribers must be able to change subscription addresses or unsubscribe at will • Human (or automated) list management removes invalid or undeliverable addresses

  4. The problem of spam • Spam costs are primarily borne by recipients • Recipients pay for services to receive email • Compare with postal bulk mail, where senders assume much more of the cost • The cost of sending one email is not much different than the cost of sending 1,000 emails • Low cost of sending spam means less resistance to doing it • Disincentives are mainly social, not financial

  5. The problem of spam, cont. • Spam is becoming a significant fraction of all email processed • For November 6: • Gladstone delivers about 36,000 messages, rejects about 17,000 spams • Darkwing delivers about 35,000 messages, rejects about 4700 spams • Yes, I know many of you probably still saw spam anyway • This was not a particularly bad day; we have had days where spam rejection exceeded normal mail volume

  6. Top N spammers, November 6

  7. Why do spammers spam? • Many spammers claim to be engaging in commercial activity, but few make money • Unfortunately, they also don't lose enough money to discourage them • Some profit by selling 'email advertising services' to naïve customers • Who then find out their 'email advertising campaign' has thoroughly cheesed off their audience • The sociopathic theory of spammers • Mainly they do it because they get off on it

  8. Background: SMTP protocol • 'Simple' Mail Transfer Protocol • Standardized since about 1982 • How email is transferred between hosts • Analogies with postal mail • 'envelope' vs. message headers/body • 'envelope sender' corresponds to return address • 'envelope recipient' corresponds to recipient address • Addresses in headers do not have to correspond to envelope addresses

  9. SMTP protocol example

  10. Message headers

  11. Characteristics of spam email • Forged headers are the norm • Easy for spammer to change • Tends to obfuscate origin for naïve readers • Common avenues of spam delivery • Spam-for-hire firms with specific network locations • ISP modem or DSL pools • Unsecured SMTP relays • Unsecured HTTP proxies

  12. Why is spam hard to block? • Spammer can control envelope sender, most message contents • Spam-for-hire outfits are always registering new domains • Spammers change ISPs (usually because they're kicked off one and move to another) • Spammers are exploiting an unknown (maybe unknowable) pool of open SMTP relays and HTTP/SOCKS proxies

  13. What can we do to block spam? • Reject sender domains commonly used by spammers • Reject connections from IP address ranges under control of spammers, or of spam-friendly ISPs • Use available databases of IP addresses of open SMTP relays, open HTTP/SOCKS proxies • Some checks for valid DNS, valid header syntax • Content filtering

  14. Details of local spam blocking • Sendmail /etc/mail/access • Primarily a database of domains or specific sender addresses that are rejected if they appear in the domain name of the connecting client or in the envelope sender • cyberpromo.com REJECT rejects sanford@cyberpromo.com, webmaster@www.cyberpromo.com, etc. • friend@public.com REJECT rejects only that specific adddress • Currently ~2300 entries on darkwing/gladstone

  15. Details of local spam blocking • /etc/hosts.deny (TCP wrappers) • Used primarily to block specific IP address ranges • Could be done in /etc/mail/access, but we have a mechanism to export this to VMS PMDF • Hoping to phase out in favor of DNS blacklists • Example: • sendmail: .cyberpromo.com 192.168.1.

  16. DNS blacklists • Special use of name server software to maintain databases of IP addresses for spam rejection • IP address is looked up under a specific zone (domain); a positive result for the lookup indicates the IP address is present in the database • Example lookups for 'spamdb.org' zone: • 2.1.168.192.spamdb.org => 127.0.0.1 (IP 192.168.1.2 is considered a spam source) • 97.142.223.128.spamdb.org => not found (IP 128.223.142.97 is not)

  17. DNS blacklists used locally • MAPS RBL+ (http://mail-abuse.org/) • rbl-plus.mail-abuse.org is lookup zone • Combination of blacklists: • RBL - mainly 'career spammers' • DUL - ISP dialup IPs, mainly contributed by the ISPs themselves • RSS - open SMTP relays • Available by subscription only, we negotiated a campus-wide subscription • DUL and RSS most effective parts

  18. DNS blacklists used locally, cont. • Spamhaus SBL (http://spamhaus.org/SBL/) • sbl.spamhaus.org is lookup zone • Free for all to use (very large sites encouraged to mirror zone data) • Very extensive listing of 'career' spammers, more effective than RBL

  19. DNS blacklists used locally, cont. • Blitzed OPM (http://blitzed.org/opm/) • opm.blitzed.org is lookup zone • Lists open HTTP/SOCKS proxies • Osirusoft DNSBL (http://relays.osirusoft.com) • relays.osirusoft.com is lookup zone • Combines many other blacklists • We currently use only the open HTTP/SOCKS proxies subset (lookups return 127.0.0.9)

  20. Avoiding 'collateral damage' • DNS blacklists chosen for consistency and stability of policies • Local blocks are based on user reports • Mail is always either delivered or bounced (returned to sender) • We avoid blocking domains sending apparently legitimate mail, even if some mail is unsolicited • Persistent spam and unresponsive mangement can overcome our reluctance

  21. Content filtering • I'm not a big fan • Tends to be complicated and difficult to predict • Examines message data entirely under spammer control, so it's easy for spammers to elude • Content criteria vary from person to person, so difficult to implement well on a system-wide level • However, can catch spam that eludes previously-described blocking methods

  22. Content filtering with spamassassin • Spamassassin is a (rather large) Perl script that applies many content checks to messages • Invoked from user .procmailrc at delivery time • :0fw| /usr/local/bin/spamasassin -P:0:* ^X-Spam-Status: Yescaughtspam • Works pretty well, but has noticeable false positive rate • User customization and 'whitelisting' is recommended

More Related