Multi-Site Dual ISP Redundant Site-to-Site VPN With OSPF Failover
Multi-Site Dual ISP Redundant Site-to-Site VPN With OSPF Failover
Multi-Site Dual ISP Redundant Site-to-Site VPN With OSPF Failover
Next
move
to
IPSec
Tunnels,
(Network-‐>IPSec
Tunnels)
there
will
be
4
tunnels
for
each
remote
site
and
8
at
the
main
site.
Give
each
tunnel
a
name,
specify
the
tunnel
interface
to
be
used
for
that
tunnel
and
in
the
drop-‐down
for
IKE
Gateway
click
the
link
to
create
a
new
IKE
Gateway.
In
the
new
IKE
Gateway
window
specify
the
name,
the
physical
interface
this
tunnel
will
be
tied
to,
select
the
IP
address
in
the
drop-‐down
(optional
if
only
a
single
IP
address
is
assigned
to
that
interface),
and
specify
the
peer
address
and
pre-‐shared
key
for
this
tunnel.
After
clicking
OK
on
the
IKE
Gateway
creation
window,
select
that
newly
created
IKE
Gateway
from
the
drop-‐down
back
in
the
IPSec
Tunnel
creation
window,
then
check
the
box
for
Tunnel
Monitor,
specify
the
IP
address
for
the
tunnel
interface
on
the
other
side
of
this
tunnel
(the
tunnel
interface’s
address
on
the
peer
side),
and
select
the
default
Monitor
profile
(this
will
be
adjusted
later,
create
a
unique
Monitor
Profile
right
away
if
desired).
Do
not
specify
any
Proxy
IDs,
leave
everything
on
that
tab
blank.
Each
of
these
tunnels
will
remain
red
(down)
until
the
configuration
is
completed
and
committed
on
both
of
the
peers.
Now
move
to
Virtual
Routers,
every
site
will
have
3
virtual
routers
(no
matter
the
number
of
tunnels);
one
for
each
ISP
and
one
for
all
other
interfaces.
The
reason
for
this
is
that
in
order
to
communicate
each
ISP
connection
will
need
a
next-‐hop.
Multiple
default
gateway
routes
in
a
single
virtual
router
will
not
accomplish
this
and
traffic
originating
from
the
firewall
does
not
follow
Policy-‐Based
Forwarding
rules.
Create
each
of
the
ISP
virtual
routers,
add
the
physical
interface
of
the
firewall
that
is
connected
to
that
ISP
and
in
static
routes
add
a
default
route
for
that
ISPs
next
hop.
After
that,
create
the
third
virtual
router
and
add
all
other
interfaces
to
this
one,
including
all
tunnel
interfaces.
If
local
internet
access
is
desired
at
that
site,
add
a
default
route
pointing
to
the
virtual
router
of
the
primary
ISP
as
the
next
hop.
Then
move
to
Redistribution
Profile
and
click
Add.
Name
it,
set
priority
(1),
select
Connect,
and
then
add
all
connected
interfaces
that
should
have
their
directly
connected
address
ranges
advertised
through
OSPF
to
the
other
locations.
Optionally
create
a
secondary
redistribution
profile
with
a
priority
of
2
selecting
Static
and
specifying
static
routes
you’d
like
to
redistribute
to
other
locations
in
the
middle
column
(under
Destination).
Switch
to
the
OSPF
tab
and
click
the
selection
box
to
enable
OSPF
and
give
a
router
ID.
(Note:
this
is
NOT
an
IP
address,
though
it
is
specified
by
4
octets
separated
by
periods
just
like
an
IP
address;
generally
using
the
IP
address
of
the
device
can
simplify
troubleshooting)
Click
Add
to
add
an
OSPF
area,
give
it
an
area
ID
(for
most
small
environments
area
0.0.0.0
will
work
perfectly
fine)
and
click
on
the
Interface
tab.
Add
each
of
the
tunnel
interfaces
here,
accepting
all
of
the
default
values
except
for
the
Link
Type,
specify
p2p
for
this.
Toward
the
end
of
this
document
tuning
operations
are
covered
to
adjust
these
timers
for
faster
failover
times.
Once
all
tunnel
interfaces
are
added
click
OK
to
return
to
the
Virtual
Router
window
on
the
OSPF
tab.
Click
the
Export
Rules
tab
and
Add
the
export
rule
previously
created
to
advertise
all
connected
subnets
out
as
an
Ext-‐1
type
and
optionally
specify
a
metric
for
it
(if
no
metric
is
specified
it
will
use
the
virtual
router’s
default
metric).
(If
a
secondary
redistribution
profile
was
created
to
advertise
static
routes,
also
add
this
one
in
the
same
manner)
At
this
point,
the
configuration
to
bring
up
the
VPN
tunnels
and
the
OSPF
neighbors
is
complete.
Verify
that
a
security
rule
is
created
allowing
traffic
to
&
from
the
“vpn”
zone
for
the
desired
areas
of
the
network
at
each
location
and
Commit
the
changes.
If
all
configuration
was
completed
successfully
there
should
be
4
tunnels
at
each
remote
site
showing
green
and
8
tunnels
at
the
main
site
showing
just
the
same.
Remote
Site
1
Remote
Site
2
Main
Site
Switch
over
to
Virtual
Routers
and
select
More
Runtime
Stats
for
the
virtual
router
that
has
all
of
the
tunnel
interfaces
associated
with
it.
On
the
OSPF
tab,
select
the
Neighbor
tab;
in
each
of
the
remote
sites
there
should
be
4
neighbors
and
at
the
main
site
there
should
be
8.
If
all
is
correct
so
far,
then
moving
to
the
Routing
tab
there
should
be
routes
for
all
of
the
local
subnets
specified
in
the
redistribution
profiles
at
each
of
the
sites
with
the
flags
A
O1
indicating
that
they
are
Active
routes,
they
were
learned
via
OSPF
and
they
are
Ext-‐1
routes.
Failover
times
in
this
configuration
will
be
approximately
10-‐15
seconds,
to
decrease
this
follow
the
below
tuning
methods.
Adjust
the
Monitor
profile
–
this
will
determine
how
long
a
tunnel
interface
is
kept
alive
when
it’s
monitored
address
is
no
longer
accessible.
(Network-‐>Monitor)
Depending
on
the
stability
of
the
connections
at
each
location
this
can
be
lowered
from
the
default
of
3
second
intervals
with
a
threshold
of
5.
In
the
lab
this
is
configured
at
2
second
intervals
with
a
threshold
of
2.
At
the
very
least,
this
should
be
switched
from
the
Action
of
“wait-‐recover”
to
“fail-‐over”.
This
will
create
faster
failovers
during
outages.
Adjust
the
OSPF
timers
–
this
will
determine
reconvergence
times
when
an
interface
drops.
(Network-‐>Virtual
Routers-‐><virtual
router
with
the
OSPF
config>-‐>OSPF
–
Edit
area
0.0.0.0)
For
each
of
the
tunnel
interfaces
the
Timing
can
be
adjusted,
primarily
focusing
on
the
Hello
Interval
and
Dead
Counts
timers.
The
timers
between
each
neighbor
connection
need
to
match,
if
they
do
not
the
neighbor
will
not
come
up,
or
it
may
come
up
but
will
cause
route
flapping.
Again,
the
ability
to
tune
these
will
depend
on
the
stability
of
the
connection
at
the
particular
location
but
in
the
lab
these
are
currently
set
at
5
and
3
respectively
with
failover
times
at
3-‐4
seconds.
Adjusting
either
of
these
two
mechanisms
too
aggressively
will
cause
flapping
interfaces
and
routes
and
will
lead
to
a
very
unstable
environment;
when
tuning,
it
is
best
to
be
not
aggressive
enough
than
too
aggressive.