Exchange 2007 video review -- disaster recovery features

http://tech-informer.com/exchange-2007-video-review-disaster-recovery-features-88/
Exchange 2007 video review -- disaster recovery features

---

Closed Caption:

so if any of that me disaster recovery
pages
you know the very first one is called
local continuous replication
and the way it works is that
led senate you got your Exchange server
and
that Exchange server might have one disc
here for the operating system
and it might have a separate disk for
your Exchange database
months or this might be split up on 32
to discuss it doesn't really matter
but what local continuous replication
enable you to do
is to grad another disc on the same
server that is and what it does
is that it able 3 to replicate the
database and the transaction logs
on 32 this new disc now the benefit that
is that if your database itself but to
become corrupt
work even the entire disk with an
offline
the administrator in exchange is able
through to go through the management
console
the management console currently be
pointing through to this location here
which is the old location aware the
database and transaction logs were
and here she can easily now redirect
that
through to the alternate location which
again is on the local disk
but a separate disc from the original
and
literally within seconds exchange is
back up and running running
of the replicated database and
transaction months
from the local disk and now as I
mentioned
this is a manufacturer and requires an
administrator
to you make this change and
it's available in both visions %uh
exchange 2007 for but standard and the
Enterprise Edition
and like I explained it to form %uh
Michael up shipping
and you'll be familiar with that time if
your sequel administrator
the next home
the disaster recovery is called
clustered continuous replication
now this particular version use
and available only on the Enterprise
Edition
and Exchange 2007 and matter fact I'm
just going to
have a quick scribbling so presley's
will make it and
a little bit more into it now
let me just before we cover
clustered continuous replication I want
to touch on the
current form the a high-availability it
available
and that is by deploying a Microsoft
server
cluster we won't happen is that have
physical mental and physical not number
two and
these three physical ninth applied in
three
to you my storage area network and in my
storage area network
I have mine sayin this and my Sandisk
and this is basically where at
store the Exchange database and
transaction month
and from within my question what I do
here
is I know create virtual instanceof
exchange
and no my clients that are out on the
network
Roman talking 32
71 also the two locally
what they do is that by now talk 32 this
Metal Exchange cel no special Exchange
server right now
is currently hosted by 71 and seven and
one in 10
is actually getting the database and
transaction logs
from the start your network ca so what
happened here is that
if anything were to happen 32 said the
number one
if the motherboard fried or anything
whatsoever
before what would happen is that the
exchange services
would now stop on this particular
machine now this lay here the actual
class the latter
is able to figure that out and realizes
that makes james hasn't stopped
running on more than one and then 21 was
the one that was currently
running the same instance an exchange it
takes is instant and Exchange
and let sit by the 32 night to
tonight to is the one that's currently
actively hosting that vigilant in
exchange now the nice thing about this
is that the clients automatically
keep talking through to the same virtual
instance that this was called exchange
23
it was called exchange 23 when I was on
ninety-one
and it's still cold exchange 23 went
from a to therefore
connectivity just returns within roughly
a matter up about one minute
that's roughly how long it takes for
this flight operation to
happen that's basically situation right
now
and with conventional clustering and
and I what happened you know what's
quite well there there's to issues
the very first issue is that you require
the storage area network
and typically that means dollars so
it can be relatively expensive for some
people
and entry-level storage area network
with roughly that one terabyte of
storage
you looking at I'm as trillion dollars
roughly
you know 30,000 the 50,000
shelling roughly and
you know that that's not exactly a put a
change in some people can afford that so
they therefore
immediately kind Guinea informant high
availability for exchange
thats basically from the morn from the
two
that problem on problem too he's that
you can see here
that we've got the exchange databases
on beasties which leaves on the storage
area network
and the same thing goes for transaction
month now we said back here
that if anything were to happen 32
exchange seven and one
that the exchange instance would flip
across
no that's fine but what happens if mind
beef
here word to you get crap
well since I'm only got a single copy on
that disc
even when exchange now floats across 32
not number two
number two still wants to access that
same disc
so that for account they said because
this disc is now gone
right so what happens is that seems I've
got
a single copy
are my storage it becomes a single point
of failure
can that the current limitation with so
the question
now let's with over 32 the new model
which is
clustered continuous replication
why this thing works is that we've got
exchange
and exchange and we have
exchange installed locally
on the discs know what happens is that
we still do create
I cluster but rather than being a server
cluster
it's what's called a majority know it
set
cluster it's a different type so what
happens here is that
rather meeting mister your network what
we're going to do is going to be
replicating
from ones that have local this 32
another said local disks on a different
machine
rights therefore there's no need for the
seine so what happens now
is that we define one
Exchange server to be active and the
same way as before we actually have a
virtual instance in exchange
which the client's I'm going to be
talking to a
run talking through to the physical
night so what happens now
is that by setting this majority not
sequester up
now might database in one transaction
logs
get replicated 32 my alternate server
and therefore this becomes my passive 70
but nice thing here is that you anything
happens through to you
this Exchange Server or my database
all my transaction logs then what's
going to happen is that
the cluster lanell you able to again
like if you like the virtual instance
and here
my clients can still get access through
the same
x10 23 7 like it was before and
everything
just keeps working internally within
roughly about one minute
and or less we've got clients returning
through to know more even though I i've
had a disaster in here
so that's basically how I majority not
cross the works
what's new is and you also require
a file server or file share if you like
and this acts as something called the
witness
no i i don't really have time to go into
explained how this thing works but
for now just take it and the point that
you need
to Exchange servers and
a third system act as a witness in order
for this majority not too close to work
get a point I want to make is that since
this technology he said
using clustering then clustering
requires
windows 2003 Enterprise Edition
thats on both nights and on public that
since we're using and continued
questions replication
then that requires exchange 2007
Enterprise Edition again number
okay so that's basically it works
hopefully that's clear
as we burn described we require in the
presentation a 2003
and Exchange I wanna limitations is
that and a between these two senators
even though they can obviously be
separated
they can be geographically dispersed box
they must be on the same summit right
say
you basically need to and ensure that
you got
a telco or you know your internet
service provider of what is providing
wide area network
is able to provide a dispersed
networks such that it actually share the
same subnet
I rather than having super 7 it
originally
as I mentioned this is an automatic
final event as a poster to local
which is million and we've talked
briefly about this majority not
sequester
and meeting fileshare witness
rights leader a2 and really quite
exciting pages
I've exchange 2007 it enables us
to you has a high-availability either
locally
with this model or a between separate
said its
with this model here and again there's
no need for the expensive storage area
network
so the cost savings he can be quite
significant
another thing to mention is that if we
look at this
scenario here we've got 17 here and got
another set here
now all over the exchange databases and
transaction logs
I continuously being replicated say
you know second by second so what
happens now is that if
a disaster happens out here well we we
have this ever that's completely up to
date
so therefore what happens is that we
know
can reduce the amount %uh tape backups
that we do
since we've got a system either here
that is effectively after the second
so I rather if you have previously doing
same daily for backups
then you might want to reconsider and
perhaps the weekly
for backups and sent again you've got
this seven to 10 here
so that can be a quite a significant
benefit as well
a secondly we've got in 2007 the
dumpster so did not have this before if
I'm you're not look and
you got your inbox and in here you've
got messages
now if you delete this particular
message
it obsolete and gets put in three-to-two
leader tons
now if he does n you delete your deleted
items
what happens now is that the Exchange
server
keeps a copy %uh these messages that you
deleted
for a period of time now that they're is
called the dumpster
in exchange 2003 the dumpster into those
three games
in exchange 2007 it's been increased
three to fourteen
said between local and and I'm
quite a continuous replication and the
fact that the dumpster has been
increased the degree of protection
that's been
and put in 32 exchange 2007 significant
and this is probably at a one my
favorite pages
an