Filtering is not a good solution in monetary terms. If you have to filter, you are paying for that spam anyway. Your ISP could offer you a better solution by allowing you to run your own mail server instead of using theirs. With respect to blocking spam at source, ISPs like UUnet do nothing. blocking them until they clean up their act seems to the best solution available currently. Which is why I use SPEWS on my mail server.
Hmmmm, Mumbai was the most expensive real estate a few years ago. If you want to compare with Manhattan, Cuffe Parade/Colaba in Mumbai start at about 2500 USD/month without parking space. Double the rental if you want parking space for a small car. Of course, by the standards of peoples income, Indian cities tend to get really expensive.
Huh? Indian cities are *expensive*. Mumbai rates are about 25K INR/sq ft. in the city to about 2K to 4K/sqft. in the suburbs. Bangalore is between 1600 to 2500 INR/sqft. Delhi is between 2K to 10K. Hyderabad is from 800 to 2500 INR/sqft.
I pay about 5500 INR/mth as rent in a fairly posh area in Bangalore. A comfortable salary would be about 30K INR. 2009 salary would be about 45 to 50K for an equivalent lifestyle.
From what I know of the people in Bangalore, there is a whole bunch of kids just out of school, and a lot more prefessionals who have over 5 years of experience in programming, with a lot of US returned professionals.
Do the world a favour and work out the costs of spam filtering. Heres a hint: Filters are CPU hungry. Mail servers generally don't have the CPU to spare for filtering too much. Spam eats bandwidth. rbls stop the connection *before* the crap comes in. Spam eats up disk space. Filters need admin time. Bayesian is per user, not global. Factor those costs in, and I can show you a *massive* increase in costs per user. And the larger your userbase, the higher the increase.
And who exactly is paying for the systems needed to handle that huge inflow of mail, parsing it, tagging it and then delivering to a users mailbox? Either accept and deliver, or just REJECT the mail.
Tagging the mail by the ISP is useless, and just a waste of money and resources.
Modern? If you ignore the newest movie, the original story is nothing like Romeo and Juliet. Its about a man who is torn between two women, the one he is married to, and the one he loves. Read the book first:).
Its called Challenge Response, and there is plenty of documentation on why this is a really bad idea. Just hunt it down in google
Re:How Do you Kow What's Right?
on
Who Is An ISP?
·
· Score: 1
What happens if I call/usr/sbin/sendmail directly from the command line? Anyway, the way a TCP client is defined is the system that sends the first syn. The server is the system that accepts and responds to the syn with a syn+ack. Follow that logic.
Re:How Do you Kow What's Right?
on
Who Is An ISP?
·
· Score: 1
But mail *servers* accept mail. Mail clients send the mail out (which may or may not be mail user agents). Except perhaps in the case of a monolithic MTA like Sendmail or Exim (which behaves as a client when sending mail to another server). Enjoy.
Mail serving is not CPU intensive. Which is why mail servers get low end CPUs and fast disk subsystems. If you want the server to tag spam, then you need a system with a fast CPU (or multiple such systems), which is expensive. Verifying the sender address in the original is far more efficient at stopping such undeliverable bounces, because the bounce MUST go to the envelop sender who exists and accepts mail.
Slashdot Mirror
Filtering is not a good solution in monetary terms. If you have to filter, you are paying for that spam anyway. Your ISP could offer you a better solution by allowing you to run your own mail server instead of using theirs.
With respect to blocking spam at source, ISPs like UUnet do nothing. blocking them until they clean up their act seems to the best solution available currently.
Which is why I use SPEWS on my mail server.
No. Spam is Unsolicited Bulk Email. Its all about consent, not about content.
(What about an unsolicted request for a vote from a political party?)
City Zip
Colombo is the capital of Sri Lanka
Hmmmm, Mumbai was the most expensive real estate a few years ago. If you want to compare with Manhattan, Cuffe Parade/Colaba in Mumbai start at about 2500 USD/month without parking space. Double the rental if you want parking space for a small car. Of course, by the standards of peoples income, Indian cities tend to get really expensive.
http://www.indlinux.org/
One of them.
Huh? Indian cities are *expensive*. /sqft. in the suburbs.
Mumbai rates are about 25K INR/sq ft. in the city to about 2K to 4K
Bangalore is between 1600 to 2500 INR/sqft.
Delhi is between 2K to 10K.
Hyderabad is from 800 to 2500 INR/sqft.
I pay about 5500 INR/mth as rent in a fairly posh area in Bangalore. A comfortable salary would be about 30K INR. 2009 salary would be about 45 to 50K for an equivalent lifestyle.
No. Those days were over two years ago. Now you need a degree, preferably in CS or Electronics, and some experience. Jobs are not easy to get.
From what I know of the people in Bangalore, there is a whole bunch of kids just out of school, and a lot more prefessionals who have over 5 years of experience in programming, with a lot of US returned professionals.
Ack! I need to preview before I post.
The word is "would". (and perlu stands for perl untrusted).
How hard woiuld it be to do this in PostgreSQL with pl/perlu and the Perl XML modules?
It does do forms, reports and a few more things.
Have you looked for pgaccess?
Its an Access like frontend to Pg. Needs Tk.
Assuming, of course, that they have opted in first.
Do the world a favour and work out the costs of spam filtering.
Heres a hint: Filters are CPU hungry. Mail servers generally don't have the CPU to spare for filtering too much. Spam eats bandwidth. rbls stop the connection *before* the crap comes in. Spam eats up disk space. Filters need admin time. Bayesian is per user, not global.
Factor those costs in, and I can show you a *massive* increase in costs per user. And the larger your userbase, the higher the increase.
Spam is Unsolicited Bulk Email.
Spam is about CONSENT, not about CONTENT.
This means that the only "correct" method of stopping spam is to reject it before the smtp data phase.
ISPs are /not/ common carriers.
Nope. Geeks making $$$ seem to be attractive, and there are plenty of geek girls here as well.
And who exactly is paying for the systems needed to handle that huge inflow of mail, parsing it, tagging it and then delivering to a users mailbox?
Either accept and deliver, or just REJECT the mail.
Tagging the mail by the ISP is useless, and just a waste of money and resources.
Modern? If you ignore the newest movie, the original story is nothing like Romeo and Juliet. Its about a man who is torn between two women, the one he is married to, and the one he loves. :).
Read the book first
Its called Challenge Response, and there is plenty of documentation on why this is a really bad idea.
Just hunt it down in google
What happens if I call /usr/sbin/sendmail directly from the command line?
Anyway, the way a TCP client is defined is the system that sends the first syn. The server is the system that accepts and responds to the syn with a syn+ack.
Follow that logic.
But mail *servers* accept mail. Mail clients send the mail out (which may or may not be mail user agents). Except perhaps in the case of a monolithic MTA like Sendmail or Exim (which behaves as a client when sending mail to another server).
Enjoy.
#!/usr/bin/perl -w
use strict;
#Print the geek way
while(1) {
print "I will never bitch about my bandwidth use again.\n";
}
No, those are .-bombs
Mail serving is not CPU intensive. Which is why mail servers get low end CPUs and fast disk subsystems.
If you want the server to tag spam, then you need a system with a fast CPU (or multiple such systems), which is expensive.
Verifying the sender address in the original is far more efficient at stopping such undeliverable bounces, because the bounce MUST go to the envelop sender who exists and accepts mail.