The document discusses the challenges of email spam filtering with IPv6. It notes that while IPv4 allows for around 4 billion IP addresses, IPv6 allows for up to 18 quintillion addresses, vastly increasing the scale of potential spamming IPs. It argues that building IP blacklists and whitelists at this scale would result in files too large to efficiently process and replicate across networks in real time. It proposes a short term solution of using smaller "allow lists" of trusted IPv6 senders rather than blocklists to help address the problem of scale posed by IPv6 for email spam filtering.
The document discusses the challenges of email spam filtering with IPv6. It notes that while IPv4 allows for around 4 billion IP addresses, IPv6 allows for up to 18 quintillion addresses, vastly increasing the scale of potential spamming IPs. It argues that building IP blacklists and whitelists at this scale would result in files too large to efficiently process and replicate across networks in real time. It proposes a short term solution of using smaller "allow lists" of trusted IPv6 senders rather than blocklists to help address the problem of scale posed by IPv6 for email spam filtering.
The document discusses the challenges of email spam filtering with IPv6. It notes that while IPv4 allows for around 4 billion IP addresses, IPv6 allows for up to 18 quintillion addresses, vastly increasing the scale of potential spamming IPs. It argues that building IP blacklists and whitelists at this scale would result in files too large to efficiently process and replicate across networks in real time. It proposes a short term solution of using smaller "allow lists" of trusted IPv6 senders rather than blocklists to help address the problem of scale posed by IPv6 for email spam filtering.
The document discusses the challenges of email spam filtering with IPv6. It notes that while IPv4 allows for around 4 billion IP addresses, IPv6 allows for up to 18 quintillion addresses, vastly increasing the scale of potential spamming IPs. It argues that building IP blacklists and whitelists at this scale would result in files too large to efficiently process and replicate across networks in real time. It proposes a short term solution of using smaller "allow lists" of trusted IPv6 senders rather than blocklists to help address the problem of scale posed by IPv6 for email spam filtering.
Modern
filters
The
problem
of
scale
• IP
lists
must
be
updated
in
(near)
real
<me
• Max
IPv4
addresses
~
4
billion
38
• Max
IPv6
addresses
~
18
x
10
– Even
if
everyone
gets
their
own
/64,
max
18
addresses
=
18
x
10
The
problem
of
scale
(cont’d)
• What
happens
when
we
build
a
beMer
spammer?
– Every
spam
comes
from
a
unique
IP
(or
limited
reuse)
– 5
billion
spamming
IPs
per
day
Size
of
file
=
190
GB
(XBL+SBL+PBL
=
138
MB)
– This
is
too
big!
Too
big
to
process
• Geo-‐distributed
systems
must
replicate
across
network
quickly
(large
files
take
too
long
for
real
<me
effec<veness).
• Processing
the
file
takes
a
long
<me.
• IP
stats
history
tables
(e.g.,
Microso^
maintains
its
own
IP
reputa<on
tables)
grows
too
big
for
so
many
unique
IPs.
Drop
in
effec<veness
If
spammers
don’t
reuse
IP
addresses,
it
makes
IP
blocklists
useless
Proposal
(short
term)
–
Allow
Lists
Allow
List
=
“I
some<mes
send
legi<mate
mail
over
IPv6.”
You
s<ll
perform
content
filtering.
We
already
do
this
for
big
mailers
like
Hotmail,
Gmail,
etc.
Either
a
central
reputa<on
service
for
IPv6,
or
build
your
own.
Do
not
allow
anyone
to
send
you
IPv6
email!
Allow
lists
are
way
smaller
and
easier
to
maintain.
Qmail Quickstarter: Install, Set Up and Run your own Email Server: A fast-paced and easy-to-follow, step-by-step guide that gets you up and running quickly