replication topology

  • Thread starter Thread starter bystander
  • Start date Start date
B

bystander

Ok, in AD sites and services our org has a hub configuration, where one main
site is the hub site, and all the other sites spoke of it. (there is a site
link for each spoked site back to the hub configured to replicate every 60
minutes)

The other spoke sites have either 1 or 2 dc's. Also these sites DC's have a
connection object for the main DC at the hub site. (configured by the KCC I
assume)

Now, everythingworks fine, no replication issues at all. My question is, if
the network link broke to the main HUB site, would the KCC figure this out
and build new connections to the other sites in order continue replication.
Or does anything else need to be done for this to happen?
 
bystander said:
Ok, in AD sites and services our org has a hub configuration, where one
main
site is the hub site, and all the other sites spoke of it. (there is a
site
link for each spoked site back to the hub configured to replicate every 60
minutes)

Those Site Links make sense.
The other spoke sites have either 1 or 2 dc's. Also these sites DC's have
a
connection object for the main DC at the hub site. (configured by the KCC
I
assume)

The KCC should do this. It should have picked one "BridgeHead
Server" (DC) in each site to actually 'connect' to the BridgeHead
Server in the Main-Hub site. (There should actually be TWO
connection objects between them though because connections are
each ONE way.)
Now, everythingworks fine, no replication issues at all. My question is,
if
the network link broke to the main HUB site, would the KCC figure this out
and build new connections to the other sites in order continue
replication.

Well, yes, and no. It would figure out that it had lost connection
to the HQ-HUB site but then it would not be able to connect to
"other sites" (you didn't mention any WANS around or across the
rim) and so nothing would replicate -- this is of course not an
actual issue if you have no other WAN connections.

On the other hand, were the HUB DCs to merely go down (while
the WAN routers continued to work) then the KCC would figure
out that it could use "transitive nature of the site links" to connect
Rim-Rim sites in pairs -- AS LONG as the WANS and routing
still work.

The automatic feature that makes the site links transitive and
makes this work is (mis)named "Site Link Bridging" (It should
be called "Site Link Grouping" or "Transitive Site Link Groups".
Or does anything else need to be done for this to happen?

How would you expect it to replicate if the WAN is down?
 
Herb Martin said:
Those Site Links make sense.


The KCC should do this. It should have picked one "BridgeHead
Server" (DC) in each site to actually 'connect' to the BridgeHead
Server in the Main-Hub site. (There should actually be TWO
connection objects between them though because connections are
each ONE way.)

There's a connection object on both the spoke site DC(s) and the HUB sites
DC (which also holds all the FSMO roles)

Well, yes, and no. It would figure out that it had lost connection
to the HQ-HUB site but then it would not be able to connect to
"other sites" (you didn't mention any WANS around or across the
rim) and so nothing would replicate -- this is of course not an
actual issue if you have no other WAN connections.

On the other hand, were the HUB DCs to merely go down (while
the WAN routers continued to work) then the KCC would figure
out that it could use "transitive nature of the site links" to connect
Rim-Rim sites in pairs -- AS LONG as the WANS and routing
still work.

Sorry, I didn't do a good job of asking that question . I'll try again : )
(but thanks for the answers) BTW the betwork is fully routed.

a. If the HUB sites WAN link was down, would the KCC make links to the
'other' sites for replication to/from each other?

b. If the HUB sites DC was down/unavailable, would the KCC create links
to/from other sites for replication?
The automatic feature that makes the site links transitive and
makes this work is (mis)named "Site Link Bridging" (It should
be called "Site Link Grouping" or "Transitive Site Link Groups".


How would you expect it to replicate if the WAN is down?

I assumed (wrongly as it will transpire) that the KCC would dynamicly build
links between the other sites for replication due to the HUB site not being
available. If this is not the case, then what is recommended practice in
the event of a WAN failure (albeit they often arent 'down' for extended
periods)? (When DC's have failed in the past I've manually triggered the
KCC on each DC Object in ADS&S)

--
Herb Martin, MCSE, MVP
Accelerated MCSE
http://www.LearnQuick.Com
[phone number on web site]
 
Sorry for this post, forgot to add it to the previous one.

The site links have a replication frequency of 60 minutes, with an open time
window in which to do this e.g. no times are specified on the site link,
it's 'open all hours'

On the connection objects however, they appear to be configured to allow
replication to occur within a 15 minute time window once every 60 minutes,
can you let me know a couple of things please?

a.which would take priority, the site link, or the connection object

b.and why would the KCC configure itself in this way.?


Herb Martin said:
bystander said:
Ok, in AD sites and services our org has a hub configuration, where one
main
site is the hub site, and all the other sites spoke of it. (there is a
site
link for each spoked site back to the hub configured to replicate every 60
minutes)

Those Site Links make sense.
The other spoke sites have either 1 or 2 dc's. Also these sites DC's have
a
connection object for the main DC at the hub site. (configured by the KCC
I
assume)

The KCC should do this. It should have picked one "BridgeHead
Server" (DC) in each site to actually 'connect' to the BridgeHead
Server in the Main-Hub site. (There should actually be TWO
connection objects between them though because connections are
each ONE way.)
Now, everythingworks fine, no replication issues at all. My question is,
if
the network link broke to the main HUB site, would the KCC figure this out
and build new connections to the other sites in order continue
replication.

Well, yes, and no. It would figure out that it had lost connection
to the HQ-HUB site but then it would not be able to connect to
"other sites" (you didn't mention any WANS around or across the
rim) and so nothing would replicate -- this is of course not an
actual issue if you have no other WAN connections.

On the other hand, were the HUB DCs to merely go down (while
the WAN routers continued to work) then the KCC would figure
out that it could use "transitive nature of the site links" to connect
Rim-Rim sites in pairs -- AS LONG as the WANS and routing
still work.

The automatic feature that makes the site links transitive and
makes this work is (mis)named "Site Link Bridging" (It should
be called "Site Link Grouping" or "Transitive Site Link Groups".
Or does anything else need to be done for this to happen?

How would you expect it to replicate if the WAN is down?


--
Herb Martin, MCSE, MVP
Accelerated MCSE
http://www.LearnQuick.Com
[phone number on web site]
 
Sorry, I didn't do a good job of asking that question . I'll try again
: )
(but thanks for the answers) BTW the betwork is fully routed.

[No one knows for sure what "fully routed" means.]

Actually, in spite of that I answered it both ways anyway.

Just go back and re-read my response for the answer to both
questions....
a. If the HUB sites WAN link was down, would the KCC make links to the
'other' sites for replication to/from each other?

If these are the ONLY WAN links then: No.

If there are alternative WAN Links then: YES.

The KCC does NOT see the actual WAN links, but only
cares are Sites and SiteLinks and if the network connectivity
operates.
b. If the HUB sites DC was down/unavailable, would the KCC create links
to/from other sites for replication?

Yes. Even with no other WAN links except Hub-Spoke the
KCC would treat the Site Links as transitivie Spoke-Hub-Spoke
(effectively) then as long as the WANS work it will bypass
the now empty intervening site.



--
Herb Martin, MCSE, MVP
Accelerated MCSE
http://www.LearnQuick.Com
[phone number on web site]

bystander said:
Herb Martin said:
Those Site Links make sense.


The KCC should do this. It should have picked one "BridgeHead
Server" (DC) in each site to actually 'connect' to the BridgeHead
Server in the Main-Hub site. (There should actually be TWO
connection objects between them though because connections are
each ONE way.)

There's a connection object on both the spoke site DC(s) and the HUB sites
DC (which also holds all the FSMO roles)

Well, yes, and no. It would figure out that it had lost connection
to the HQ-HUB site but then it would not be able to connect to
"other sites" (you didn't mention any WANS around or across the
rim) and so nothing would replicate -- this is of course not an
actual issue if you have no other WAN connections.

On the other hand, were the HUB DCs to merely go down (while
the WAN routers continued to work) then the KCC would figure
out that it could use "transitive nature of the site links" to connect
Rim-Rim sites in pairs -- AS LONG as the WANS and routing
still work.

Sorry, I didn't do a good job of asking that question . I'll try again
: )
(but thanks for the answers) BTW the betwork is fully routed.

a. If the HUB sites WAN link was down, would the KCC make links to the
'other' sites for replication to/from each other?

b. If the HUB sites DC was down/unavailable, would the KCC create links
to/from other sites for replication?
The automatic feature that makes the site links transitive and
makes this work is (mis)named "Site Link Bridging" (It should
be called "Site Link Grouping" or "Transitive Site Link Groups".


How would you expect it to replicate if the WAN is down?

I assumed (wrongly as it will transpire) that the KCC would dynamicly
build
links between the other sites for replication due to the HUB site not
being
available. If this is not the case, then what is recommended practice in
the event of a WAN failure (albeit they often arent 'down' for extended
periods)? (When DC's have failed in the past I've manually triggered the
KCC on each DC Object in ADS&S)

--
Herb Martin, MCSE, MVP
Accelerated MCSE
http://www.LearnQuick.Com
[phone number on web site]
 
Sorry for this post, forgot to add it to the previous one.

The site links have a replication frequency of 60 minutes, with an open time
window in which to do this e.g. no times are specified on the site link,
it's 'open all hours'

On the connection objects however, they appear to be configured to allow
replication to occur within a 15 minute time window once every 60 minutes,
can you let me know a couple of things please?

a.which would take priority, the site link, or the connection object

b.and why would the KCC configure itself in this way.?

Herb Martin said:
Sorry, I didn't do a good job of asking that question . I'll try again
: )
(but thanks for the answers) BTW the betwork is fully routed.

[No one knows for sure what "fully routed" means.]

Actually, in spite of that I answered it both ways anyway.

Just go back and re-read my response for the answer to both
questions....
a. If the HUB sites WAN link was down, would the KCC make links to the
'other' sites for replication to/from each other?

If these are the ONLY WAN links then: No.

If there are alternative WAN Links then: YES.

The KCC does NOT see the actual WAN links, but only
cares are Sites and SiteLinks and if the network connectivity
operates.
b. If the HUB sites DC was down/unavailable, would the KCC create links
to/from other sites for replication?

Yes. Even with no other WAN links except Hub-Spoke the
KCC would treat the Site Links as transitivie Spoke-Hub-Spoke
(effectively) then as long as the WANS work it will bypass
the now empty intervening site.



--
Herb Martin, MCSE, MVP
Accelerated MCSE
http://www.LearnQuick.Com
[phone number on web site]

bystander said:
Herb Martin said:
"bystander" <;';';> wrote in message
Ok, in AD sites and services our org has a hub configuration, where one
main
site is the hub site, and all the other sites spoke of it. (there is a
site
link for each spoked site back to the hub configured to replicate
every
60
minutes)

Those Site Links make sense.

The other spoke sites have either 1 or 2 dc's. Also these sites DC's have
a
connection object for the main DC at the hub site. (configured by the KCC
I
assume)

The KCC should do this. It should have picked one "BridgeHead
Server" (DC) in each site to actually 'connect' to the BridgeHead
Server in the Main-Hub site. (There should actually be TWO
connection objects between them though because connections are
each ONE way.)

There's a connection object on both the spoke site DC(s) and the HUB sites
DC (which also holds all the FSMO roles)

Now, everythingworks fine, no replication issues at all. My question is,
if
the network link broke to the main HUB site, would the KCC figure
this
out
and build new connections to the other sites in order continue
replication.

Well, yes, and no. It would figure out that it had lost connection
to the HQ-HUB site but then it would not be able to connect to
"other sites" (you didn't mention any WANS around or across the
rim) and so nothing would replicate -- this is of course not an
actual issue if you have no other WAN connections.

On the other hand, were the HUB DCs to merely go down (while
the WAN routers continued to work) then the KCC would figure
out that it could use "transitive nature of the site links" to connect
Rim-Rim sites in pairs -- AS LONG as the WANS and routing
still work.

Sorry, I didn't do a good job of asking that question . I'll try again
: )
(but thanks for the answers) BTW the betwork is fully routed.

a. If the HUB sites WAN link was down, would the KCC make links to the
'other' sites for replication to/from each other?

b. If the HUB sites DC was down/unavailable, would the KCC create links
to/from other sites for replication?
The automatic feature that makes the site links transitive and
makes this work is (mis)named "Site Link Bridging" (It should
be called "Site Link Grouping" or "Transitive Site Link Groups".

Or does anything else need to be done for this to happen?

How would you expect it to replicate if the WAN is down?

I assumed (wrongly as it will transpire) that the KCC would dynamicly
build
links between the other sites for replication due to the HUB site not
being
available. If this is not the case, then what is recommended practice in
the event of a WAN failure (albeit they often arent 'down' for extended
periods)? (When DC's have failed in the past I've manually triggered the
KCC on each DC Object in ADS&S)

--
Herb Martin, MCSE, MVP
Accelerated MCSE
http://www.LearnQuick.Com
[phone number on web site]
 
bystander said:
Sorry for this post, forgot to add it to the previous one.

The site links have a replication frequency of 60 minutes, with an open
time
window in which to do this e.g. no times are specified on the site link,
it's 'open all hours'

On the connection objects however, they appear to be configured to allow
replication to occur within a 15 minute time window once every 60 minutes,
can you let me know a couple of things please?

Connection object takes precedence SINCE it must be
MANUAL to even be different.

But ONLY Schedule (times of day) not frequency is even
settable on a connection object.
a.which would take priority, the site link, or the connection object
b.and why would the KCC configure itself in this way.?

It would not. The KCC will create the connection objects
BASED on the values of the SiteLink ONLY. (Schedule
times, at Frequency intervals, and using the cost to decide
WHICH Connection objects to create when there are
choices.)

One reason to create a manual connection object (there are
seldom very good reasons) is to override that Schedule but
frequency cannot be set.

Connection objects created by the KCC have the 'name'
<automatically generated> in the name column of Sites
and Services for that object.

--
Herb Martin, MCSE, MVP
Accelerated MCSE
http://www.LearnQuick.Com
[phone number on web site]
Herb Martin said:
Sorry, I didn't do a good job of asking that question . I'll try again
: )
(but thanks for the answers) BTW the betwork is fully routed.

[No one knows for sure what "fully routed" means.]

Actually, in spite of that I answered it both ways anyway.

Just go back and re-read my response for the answer to both
questions....
a. If the HUB sites WAN link was down, would the KCC make links to the
'other' sites for replication to/from each other?

If these are the ONLY WAN links then: No.

If there are alternative WAN Links then: YES.

The KCC does NOT see the actual WAN links, but only
cares are Sites and SiteLinks and if the network connectivity
operates.
b. If the HUB sites DC was down/unavailable, would the KCC create links
to/from other sites for replication?

Yes. Even with no other WAN links except Hub-Spoke the
KCC would treat the Site Links as transitivie Spoke-Hub-Spoke
(effectively) then as long as the WANS work it will bypass
the now empty intervening site.



--
Herb Martin, MCSE, MVP
Accelerated MCSE
http://www.LearnQuick.Com
[phone number on web site]

bystander said:
"bystander" <;';';> wrote in message
Ok, in AD sites and services our org has a hub configuration, where one
main
site is the hub site, and all the other sites spoke of it. (there is a
site
link for each spoked site back to the hub configured to replicate every
60
minutes)

Those Site Links make sense.

The other spoke sites have either 1 or 2 dc's. Also these sites DC's
have
a
connection object for the main DC at the hub site. (configured by
the
KCC
I
assume)

The KCC should do this. It should have picked one "BridgeHead
Server" (DC) in each site to actually 'connect' to the BridgeHead
Server in the Main-Hub site. (There should actually be TWO
connection objects between them though because connections are
each ONE way.)

There's a connection object on both the spoke site DC(s) and the HUB sites
DC (which also holds all the FSMO roles)


Now, everythingworks fine, no replication issues at all. My
question
is,
if
the network link broke to the main HUB site, would the KCC figure this
out
and build new connections to the other sites in order continue
replication.

Well, yes, and no. It would figure out that it had lost connection
to the HQ-HUB site but then it would not be able to connect to
"other sites" (you didn't mention any WANS around or across the
rim) and so nothing would replicate -- this is of course not an
actual issue if you have no other WAN connections.

On the other hand, were the HUB DCs to merely go down (while
the WAN routers continued to work) then the KCC would figure
out that it could use "transitive nature of the site links" to connect
Rim-Rim sites in pairs -- AS LONG as the WANS and routing
still work.

Sorry, I didn't do a good job of asking that question . I'll try again
: )
(but thanks for the answers) BTW the betwork is fully routed.

a. If the HUB sites WAN link was down, would the KCC make links to the
'other' sites for replication to/from each other?

b. If the HUB sites DC was down/unavailable, would the KCC create links
to/from other sites for replication?


The automatic feature that makes the site links transitive and
makes this work is (mis)named "Site Link Bridging" (It should
be called "Site Link Grouping" or "Transitive Site Link Groups".

Or does anything else need to be done for this to happen?

How would you expect it to replicate if the WAN is down?

I assumed (wrongly as it will transpire) that the KCC would dynamicly
build
links between the other sites for replication due to the HUB site not
being
available. If this is not the case, then what is recommended practice in
the event of a WAN failure (albeit they often arent 'down' for extended
periods)? (When DC's have failed in the past I've manually triggered the
KCC on each DC Object in ADS&S)




--
Herb Martin, MCSE, MVP
Accelerated MCSE
http://www.LearnQuick.Com
[phone number on web site]
 
Herb - I thankyou for all your help with this in clarifying things for me.

Just so I am 100% sure I understand, can I run over one previous thing
please, for my own benefit.

OK, so as I've understood the above, the HUB site has 1 WAN link out to the
rest of the world, if that breaks replication will stop, configured as it
is?

Thanks, Andy



Herb Martin said:
bystander said:
Sorry for this post, forgot to add it to the previous one.

The site links have a replication frequency of 60 minutes, with an open
time
window in which to do this e.g. no times are specified on the site link,
it's 'open all hours'

On the connection objects however, they appear to be configured to allow
replication to occur within a 15 minute time window once every 60 minutes,
can you let me know a couple of things please?

Connection object takes precedence SINCE it must be
MANUAL to even be different.

But ONLY Schedule (times of day) not frequency is even
settable on a connection object.
a.which would take priority, the site link, or the connection object
b.and why would the KCC configure itself in this way.?

It would not. The KCC will create the connection objects
BASED on the values of the SiteLink ONLY. (Schedule
times, at Frequency intervals, and using the cost to decide
WHICH Connection objects to create when there are
choices.)

One reason to create a manual connection object (there are
seldom very good reasons) is to override that Schedule but
frequency cannot be set.

Connection objects created by the KCC have the 'name'
<automatically generated> in the name column of Sites
and Services for that object.

--
Herb Martin, MCSE, MVP
Accelerated MCSE
http://www.LearnQuick.Com
[phone number on web site]
Herb Martin said:
Sorry, I didn't do a good job of asking that question . I'll try again
: )
(but thanks for the answers) BTW the betwork is fully routed.

[No one knows for sure what "fully routed" means.]

Actually, in spite of that I answered it both ways anyway.

Just go back and re-read my response for the answer to both
questions....

a. If the HUB sites WAN link was down, would the KCC make links to the
'other' sites for replication to/from each other?

If these are the ONLY WAN links then: No.

If there are alternative WAN Links then: YES.

The KCC does NOT see the actual WAN links, but only
cares are Sites and SiteLinks and if the network connectivity
operates.

b. If the HUB sites DC was down/unavailable, would the KCC create links
to/from other sites for replication?

Yes. Even with no other WAN links except Hub-Spoke the
KCC would treat the Site Links as transitivie Spoke-Hub-Spoke
(effectively) then as long as the WANS work it will bypass
the now empty intervening site.



--
Herb Martin, MCSE, MVP
Accelerated MCSE
http://www.LearnQuick.Com
[phone number on web site]

"bystander" <;';';> wrote in message
"bystander" <;';';> wrote in message
Ok, in AD sites and services our org has a hub configuration,
where
one
main
site is the hub site, and all the other sites spoke of it. (there
is
a
site
link for each spoked site back to the hub configured to replicate every
60
minutes)

Those Site Links make sense.

The other spoke sites have either 1 or 2 dc's. Also these sites DC's
have
a
connection object for the main DC at the hub site. (configured by
the
KCC
I
assume)

The KCC should do this. It should have picked one "BridgeHead
Server" (DC) in each site to actually 'connect' to the BridgeHead
Server in the Main-Hub site. (There should actually be TWO
connection objects between them though because connections are
each ONE way.)

There's a connection object on both the spoke site DC(s) and the HUB sites
DC (which also holds all the FSMO roles)


Now, everythingworks fine, no replication issues at all. My
question
is,
if
the network link broke to the main HUB site, would the KCC figure this
out
and build new connections to the other sites in order continue
replication.

Well, yes, and no. It would figure out that it had lost connection
to the HQ-HUB site but then it would not be able to connect to
"other sites" (you didn't mention any WANS around or across the
rim) and so nothing would replicate -- this is of course not an
actual issue if you have no other WAN connections.

On the other hand, were the HUB DCs to merely go down (while
the WAN routers continued to work) then the KCC would figure
out that it could use "transitive nature of the site links" to connect
Rim-Rim sites in pairs -- AS LONG as the WANS and routing
still work.

Sorry, I didn't do a good job of asking that question . I'll try again
: )
(but thanks for the answers) BTW the betwork is fully routed.

a. If the HUB sites WAN link was down, would the KCC make links to the
'other' sites for replication to/from each other?

b. If the HUB sites DC was down/unavailable, would the KCC create links
to/from other sites for replication?


The automatic feature that makes the site links transitive and
makes this work is (mis)named "Site Link Bridging" (It should
be called "Site Link Grouping" or "Transitive Site Link Groups".

Or does anything else need to be done for this to happen?

How would you expect it to replicate if the WAN is down?

I assumed (wrongly as it will transpire) that the KCC would dynamicly
build
links between the other sites for replication due to the HUB site not
being
available. If this is not the case, then what is recommended
practice
in
the event of a WAN failure (albeit they often arent 'down' for extended
periods)? (When DC's have failed in the past I've manually triggered the
KCC on each DC Object in ADS&S)




--
Herb Martin, MCSE, MVP
Accelerated MCSE
http://www.LearnQuick.Com
[phone number on web site]
 
bystander said:
Herb - I thankyou for all your help with this in clarifying things for me.

We help as much as we are able.
Just so I am 100% sure I understand, can I run over one previous thing
please, for my own benefit.

See below...
OK, so as I've understood the above, the HUB site has 1 WAN link out to
the
rest of the world, if that breaks replication will stop, configured as it
is?

Yes, but this is merely simple routing. If the WAN is down
then there is NO IP traffic at all and so anything, including
AD replication, that depends on IP (and the underlying WAN)
is off until the WAN is repaired.

AD might replicate 'around' such problems if there were
additional routing paths over other WAN lines but in truth
AD really has no concept or knowledge of your routing paths
EXCEPT as you choose to describe them through adding
SitesLinks and configuring them, especially 'costs'.

One is usually wise to setup those SiteLinks to FOLLOW the
actual WAN Paths (it is nearly impossible to invent a sensible
case in which a SiteLink without an underlying WAN path
would ever make sense) but this is not however a technical
rule should someone choose do otherwise.

AD merely sets up the best "connections" it can invent (by
using "costs" to decide primarily) and then expects the
routers to deliver the actual packets.

If you have done your part correctly, i.e.,;

Sites with Subnets
SiteLinks with Cost, Schedule and Frequency

....then the KCC will do a good job of both setting up and
maintaining the "connections" between actual DCs, both
locally and by choosing a BridgeHead server (DC) in
each site to replicate with the "most adjacent" nearby
site(s) based on those costs.

Should the health of DCs change, the KCC will pick a
new Bridgehead server or even create new connections,
always attempting to maintain full replication but forced
to follow YOUR rules for Sites and Sitelinks.

Should you ever choose to create manual "connections"
then the KCC will defer to YOURS, and use them in
preference to any it would create until you remove that
manual connection.

Generally there isn't much to a hub and spoke design
following real WAN lines in the same pattern. Just
add a SiteLink between each branch site and the central
site. Set the Schedule and Frequency to something
sensible for YOU, and the cost to pretty much anything
most of the time -- since the KCC is going to prefer to
replicate direct Hub-Spoke and not Spoke-Spoke unless
all DCs in the Hub are down.

In most cases this is not much of an issue since the Hub
is almost always a "main site" where the DCs are mostly
likely to be both fault tolerant and well-monitored.

This may not be true 100% of the time but it is so common
as to be nearly always the rule.
Thanks, Andy


--
Herb Martin, MCSE, MVP
Accelerated MCSE
http://www.LearnQuick.Com
[phone number on web site]
Herb Martin said:
bystander said:
Sorry for this post, forgot to add it to the previous one.

The site links have a replication frequency of 60 minutes, with an open
time
window in which to do this e.g. no times are specified on the site
link,
it's 'open all hours'

On the connection objects however, they appear to be configured to
allow
replication to occur within a 15 minute time window once every 60 minutes,
can you let me know a couple of things please?

Connection object takes precedence SINCE it must be
MANUAL to even be different.

But ONLY Schedule (times of day) not frequency is even
settable on a connection object.
a.which would take priority, the site link, or the connection object
b.and why would the KCC configure itself in this way.?

It would not. The KCC will create the connection objects
BASED on the values of the SiteLink ONLY. (Schedule
times, at Frequency intervals, and using the cost to decide
WHICH Connection objects to create when there are
choices.)

One reason to create a manual connection object (there are
seldom very good reasons) is to override that Schedule but
frequency cannot be set.

Connection objects created by the KCC have the 'name'
<automatically generated> in the name column of Sites
and Services for that object.

--
Herb Martin, MCSE, MVP
Accelerated MCSE
http://www.LearnQuick.Com
[phone number on web site]
Sorry, I didn't do a good job of asking that question . I'll try again
: )
(but thanks for the answers) BTW the betwork is fully routed.

[No one knows for sure what "fully routed" means.]

Actually, in spite of that I answered it both ways anyway.

Just go back and re-read my response for the answer to both
questions....

a. If the HUB sites WAN link was down, would the KCC make links to the
'other' sites for replication to/from each other?

If these are the ONLY WAN links then: No.

If there are alternative WAN Links then: YES.

The KCC does NOT see the actual WAN links, but only
cares are Sites and SiteLinks and if the network connectivity
operates.

b. If the HUB sites DC was down/unavailable, would the KCC create links
to/from other sites for replication?

Yes. Even with no other WAN links except Hub-Spoke the
KCC would treat the Site Links as transitivie Spoke-Hub-Spoke
(effectively) then as long as the WANS work it will bypass
the now empty intervening site.



--
Herb Martin, MCSE, MVP
Accelerated MCSE
http://www.LearnQuick.Com
[phone number on web site]

"bystander" <;';';> wrote in message
"bystander" <;';';> wrote in message
Ok, in AD sites and services our org has a hub configuration, where
one
main
site is the hub site, and all the other sites spoke of it. (there is
a
site
link for each spoked site back to the hub configured to replicate
every
60
minutes)

Those Site Links make sense.

The other spoke sites have either 1 or 2 dc's. Also these sites DC's
have
a
connection object for the main DC at the hub site. (configured by
the
KCC
I
assume)

The KCC should do this. It should have picked one "BridgeHead
Server" (DC) in each site to actually 'connect' to the BridgeHead
Server in the Main-Hub site. (There should actually be TWO
connection objects between them though because connections are
each ONE way.)

There's a connection object on both the spoke site DC(s) and the HUB
sites
DC (which also holds all the FSMO roles)


Now, everythingworks fine, no replication issues at all. My
question
is,
if
the network link broke to the main HUB site, would the KCC figure
this
out
and build new connections to the other sites in order continue
replication.

Well, yes, and no. It would figure out that it had lost connection
to the HQ-HUB site but then it would not be able to connect to
"other sites" (you didn't mention any WANS around or across the
rim) and so nothing would replicate -- this is of course not an
actual issue if you have no other WAN connections.

On the other hand, were the HUB DCs to merely go down (while
the WAN routers continued to work) then the KCC would figure
out that it could use "transitive nature of the site links" to connect
Rim-Rim sites in pairs -- AS LONG as the WANS and routing
still work.

Sorry, I didn't do a good job of asking that question . I'll try again
: )
(but thanks for the answers) BTW the betwork is fully routed.

a. If the HUB sites WAN link was down, would the KCC make links to the
'other' sites for replication to/from each other?

b. If the HUB sites DC was down/unavailable, would the KCC create links
to/from other sites for replication?


The automatic feature that makes the site links transitive and
makes this work is (mis)named "Site Link Bridging" (It should
be called "Site Link Grouping" or "Transitive Site Link Groups".

Or does anything else need to be done for this to happen?

How would you expect it to replicate if the WAN is down?

I assumed (wrongly as it will transpire) that the KCC would
dynamicly
build
links between the other sites for replication due to the HUB site
not
being
available. If this is not the case, then what is recommended practice
in
the event of a WAN failure (albeit they often arent 'down' for extended
periods)? (When DC's have failed in the past I've manually
triggered
the
KCC on each DC Object in ADS&S)




--
Herb Martin, MCSE, MVP
Accelerated MCSE
http://www.LearnQuick.Com
[phone number on web site]
 
Herb your a star.

Thanks for all youre time (and input).


Herb Martin said:
bystander said:
Herb - I thankyou for all your help with this in clarifying things for
me.

We help as much as we are able.
Just so I am 100% sure I understand, can I run over one previous thing
please, for my own benefit.

See below...
OK, so as I've understood the above, the HUB site has 1 WAN link out to
the
rest of the world, if that breaks replication will stop, configured as it
is?

Yes, but this is merely simple routing. If the WAN is down
then there is NO IP traffic at all and so anything, including
AD replication, that depends on IP (and the underlying WAN)
is off until the WAN is repaired.

AD might replicate 'around' such problems if there were
additional routing paths over other WAN lines but in truth
AD really has no concept or knowledge of your routing paths
EXCEPT as you choose to describe them through adding
SitesLinks and configuring them, especially 'costs'.

One is usually wise to setup those SiteLinks to FOLLOW the
actual WAN Paths (it is nearly impossible to invent a sensible
case in which a SiteLink without an underlying WAN path
would ever make sense) but this is not however a technical
rule should someone choose do otherwise.

AD merely sets up the best "connections" it can invent (by
using "costs" to decide primarily) and then expects the
routers to deliver the actual packets.

If you have done your part correctly, i.e.,;

Sites with Subnets
SiteLinks with Cost, Schedule and Frequency

...then the KCC will do a good job of both setting up and
maintaining the "connections" between actual DCs, both
locally and by choosing a BridgeHead server (DC) in
each site to replicate with the "most adjacent" nearby
site(s) based on those costs.

Should the health of DCs change, the KCC will pick a
new Bridgehead server or even create new connections,
always attempting to maintain full replication but forced
to follow YOUR rules for Sites and Sitelinks.

Should you ever choose to create manual "connections"
then the KCC will defer to YOURS, and use them in
preference to any it would create until you remove that
manual connection.

Generally there isn't much to a hub and spoke design
following real WAN lines in the same pattern. Just
add a SiteLink between each branch site and the central
site. Set the Schedule and Frequency to something
sensible for YOU, and the cost to pretty much anything
most of the time -- since the KCC is going to prefer to
replicate direct Hub-Spoke and not Spoke-Spoke unless
all DCs in the Hub are down.

In most cases this is not much of an issue since the Hub
is almost always a "main site" where the DCs are mostly
likely to be both fault tolerant and well-monitored.

This may not be true 100% of the time but it is so common
as to be nearly always the rule.
Thanks, Andy


--
Herb Martin, MCSE, MVP
Accelerated MCSE
http://www.LearnQuick.Com
[phone number on web site]
Herb Martin said:
"bystander" <;';';> wrote in message
Sorry for this post, forgot to add it to the previous one.

The site links have a replication frequency of 60 minutes, with an open
time
window in which to do this e.g. no times are specified on the site
link,
it's 'open all hours'

On the connection objects however, they appear to be configured to
allow
replication to occur within a 15 minute time window once every 60 minutes,
can you let me know a couple of things please?

Connection object takes precedence SINCE it must be
MANUAL to even be different.

But ONLY Schedule (times of day) not frequency is even
settable on a connection object.

a.which would take priority, the site link, or the connection object
b.and why would the KCC configure itself in this way.?

It would not. The KCC will create the connection objects
BASED on the values of the SiteLink ONLY. (Schedule
times, at Frequency intervals, and using the cost to decide
WHICH Connection objects to create when there are
choices.)

One reason to create a manual connection object (there are
seldom very good reasons) is to override that Schedule but
frequency cannot be set.

Connection objects created by the KCC have the 'name'
<automatically generated> in the name column of Sites
and Services for that object.

--
Herb Martin, MCSE, MVP
Accelerated MCSE
http://www.LearnQuick.Com
[phone number on web site]

Sorry, I didn't do a good job of asking that question . I'll try again
: )
(but thanks for the answers) BTW the betwork is fully routed.

[No one knows for sure what "fully routed" means.]

Actually, in spite of that I answered it both ways anyway.

Just go back and re-read my response for the answer to both
questions....

a. If the HUB sites WAN link was down, would the KCC make links to the
'other' sites for replication to/from each other?

If these are the ONLY WAN links then: No.

If there are alternative WAN Links then: YES.

The KCC does NOT see the actual WAN links, but only
cares are Sites and SiteLinks and if the network connectivity
operates.

b. If the HUB sites DC was down/unavailable, would the KCC create links
to/from other sites for replication?

Yes. Even with no other WAN links except Hub-Spoke the
KCC would treat the Site Links as transitivie Spoke-Hub-Spoke
(effectively) then as long as the WANS work it will bypass
the now empty intervening site.



--
Herb Martin, MCSE, MVP
Accelerated MCSE
http://www.LearnQuick.Com
[phone number on web site]

"bystander" <;';';> wrote in message
"bystander" <;';';> wrote in message
Ok, in AD sites and services our org has a hub configuration, where
one
main
site is the hub site, and all the other sites spoke of it.
(there
is
a
site
link for each spoked site back to the hub configured to replicate
every
60
minutes)

Those Site Links make sense.

The other spoke sites have either 1 or 2 dc's. Also these sites DC's
have
a
connection object for the main DC at the hub site. (configured by
the
KCC
I
assume)

The KCC should do this. It should have picked one "BridgeHead
Server" (DC) in each site to actually 'connect' to the BridgeHead
Server in the Main-Hub site. (There should actually be TWO
connection objects between them though because connections are
each ONE way.)

There's a connection object on both the spoke site DC(s) and the HUB
sites
DC (which also holds all the FSMO roles)


Now, everythingworks fine, no replication issues at all. My
question
is,
if
the network link broke to the main HUB site, would the KCC figure
this
out
and build new connections to the other sites in order continue
replication.

Well, yes, and no. It would figure out that it had lost connection
to the HQ-HUB site but then it would not be able to connect to
"other sites" (you didn't mention any WANS around or across the
rim) and so nothing would replicate -- this is of course not an
actual issue if you have no other WAN connections.

On the other hand, were the HUB DCs to merely go down (while
the WAN routers continued to work) then the KCC would figure
out that it could use "transitive nature of the site links" to connect
Rim-Rim sites in pairs -- AS LONG as the WANS and routing
still work.

Sorry, I didn't do a good job of asking that question . I'll try again
: )
(but thanks for the answers) BTW the betwork is fully routed.

a. If the HUB sites WAN link was down, would the KCC make links to the
'other' sites for replication to/from each other?

b. If the HUB sites DC was down/unavailable, would the KCC create links
to/from other sites for replication?


The automatic feature that makes the site links transitive and
makes this work is (mis)named "Site Link Bridging" (It should
be called "Site Link Grouping" or "Transitive Site Link Groups".

Or does anything else need to be done for this to happen?

How would you expect it to replicate if the WAN is down?

I assumed (wrongly as it will transpire) that the KCC would
dynamicly
build
links between the other sites for replication due to the HUB site
not
being
available. If this is not the case, then what is recommended practice
in
the event of a WAN failure (albeit they often arent 'down' for extended
periods)? (When DC's have failed in the past I've manually
triggered
the
KCC on each DC Object in ADS&S)




--
Herb Martin, MCSE, MVP
Accelerated MCSE
http://www.LearnQuick.Com
[phone number on web site]
 
Andrew Story said:
Herb your a star.

Thanks for all youre time (and input).

You are most welcome.

Pass on the help to others, and recommend me to your friends <grin>


--
Herb Martin, MCSE, MVP
Accelerated MCSE
http://www.LearnQuick.Com
[phone number on web site]




Andrew Story said:
Herb your a star.

Thanks for all youre time (and input).


Herb Martin said:
bystander said:
Herb - I thankyou for all your help with this in clarifying things for
me.

We help as much as we are able.
Just so I am 100% sure I understand, can I run over one previous thing
please, for my own benefit.

See below...
a. If the HUB sites WAN link was down, would the KCC make links to the
'other' sites for replication to/from each other?

If these are the ONLY WAN links then: No.

If there are alternative WAN Links then: YES.

The KCC does NOT see the actual WAN links, but only
cares are Sites and SiteLinks and if the network connectivity
operates.

b. If the HUB sites DC was down/unavailable, would the KCC create
links
to/from other sites for replication?

Yes. Even with no other WAN links except Hub-Spoke the
KCC would treat the Site Links as transitivie Spoke-Hub-Spoke
(effectively) then as long as the WANS work it will bypass
the now empty intervening site.

OK, so as I've understood the above, the HUB site has 1 WAN link out to
the
rest of the world, if that breaks replication will stop, configured as it
is?

Yes, but this is merely simple routing. If the WAN is down
then there is NO IP traffic at all and so anything, including
AD replication, that depends on IP (and the underlying WAN)
is off until the WAN is repaired.

AD might replicate 'around' such problems if there were
additional routing paths over other WAN lines but in truth
AD really has no concept or knowledge of your routing paths
EXCEPT as you choose to describe them through adding
SitesLinks and configuring them, especially 'costs'.

One is usually wise to setup those SiteLinks to FOLLOW the
actual WAN Paths (it is nearly impossible to invent a sensible
case in which a SiteLink without an underlying WAN path
would ever make sense) but this is not however a technical
rule should someone choose do otherwise.

AD merely sets up the best "connections" it can invent (by
using "costs" to decide primarily) and then expects the
routers to deliver the actual packets.

If you have done your part correctly, i.e.,;

Sites with Subnets
SiteLinks with Cost, Schedule and Frequency

...then the KCC will do a good job of both setting up and
maintaining the "connections" between actual DCs, both
locally and by choosing a BridgeHead server (DC) in
each site to replicate with the "most adjacent" nearby
site(s) based on those costs.

Should the health of DCs change, the KCC will pick a
new Bridgehead server or even create new connections,
always attempting to maintain full replication but forced
to follow YOUR rules for Sites and Sitelinks.

Should you ever choose to create manual "connections"
then the KCC will defer to YOURS, and use them in
preference to any it would create until you remove that
manual connection.

Generally there isn't much to a hub and spoke design
following real WAN lines in the same pattern. Just
add a SiteLink between each branch site and the central
site. Set the Schedule and Frequency to something
sensible for YOU, and the cost to pretty much anything
most of the time -- since the KCC is going to prefer to
replicate direct Hub-Spoke and not Spoke-Spoke unless
all DCs in the Hub are down.

In most cases this is not much of an issue since the Hub
is almost always a "main site" where the DCs are mostly
likely to be both fault tolerant and well-monitored.

This may not be true 100% of the time but it is so common
as to be nearly always the rule.
Thanks, Andy


--
Herb Martin, MCSE, MVP
Accelerated MCSE
http://www.LearnQuick.Com
[phone number on web site]
"bystander" <;';';> wrote in message
Sorry for this post, forgot to add it to the previous one.

The site links have a replication frequency of 60 minutes, with an open
time
window in which to do this e.g. no times are specified on the site
link,
it's 'open all hours'

On the connection objects however, they appear to be configured to
allow
replication to occur within a 15 minute time window once every 60
minutes,
can you let me know a couple of things please?

Connection object takes precedence SINCE it must be
MANUAL to even be different.

But ONLY Schedule (times of day) not frequency is even
settable on a connection object.

a.which would take priority, the site link, or the connection object
b.and why would the KCC configure itself in this way.?

It would not. The KCC will create the connection objects
BASED on the values of the SiteLink ONLY. (Schedule
times, at Frequency intervals, and using the cost to decide
WHICH Connection objects to create when there are
choices.)

One reason to create a manual connection object (there are
seldom very good reasons) is to override that Schedule but
frequency cannot be set.

Connection objects created by the KCC have the 'name'
<automatically generated> in the name column of Sites
and Services for that object.

--
Herb Martin, MCSE, MVP
Accelerated MCSE
http://www.LearnQuick.Com
[phone number on web site]

Sorry, I didn't do a good job of asking that question . I'll try
again
: )
(but thanks for the answers) BTW the betwork is fully routed.

[No one knows for sure what "fully routed" means.]

Actually, in spite of that I answered it both ways anyway.

Just go back and re-read my response for the answer to both
questions....

a. If the HUB sites WAN link was down, would the KCC make links
to
the
'other' sites for replication to/from each other?

If these are the ONLY WAN links then: No.

If there are alternative WAN Links then: YES.

The KCC does NOT see the actual WAN links, but only
cares are Sites and SiteLinks and if the network connectivity
operates.

b. If the HUB sites DC was down/unavailable, would the KCC create
links
to/from other sites for replication?

Yes. Even with no other WAN links except Hub-Spoke the
KCC would treat the Site Links as transitivie Spoke-Hub-Spoke
(effectively) then as long as the WANS work it will bypass
the now empty intervening site.



--
Herb Martin, MCSE, MVP
Accelerated MCSE
http://www.LearnQuick.Com
[phone number on web site]

"bystander" <;';';> wrote in message
"bystander" <;';';> wrote in message
Ok, in AD sites and services our org has a hub configuration,
where
one
main
site is the hub site, and all the other sites spoke of it. (there
is
a
site
link for each spoked site back to the hub configured to replicate
every
60
minutes)

Those Site Links make sense.

The other spoke sites have either 1 or 2 dc's. Also these
sites
DC's
have
a
connection object for the main DC at the hub site. (configured by
the
KCC
I
assume)

The KCC should do this. It should have picked one "BridgeHead
Server" (DC) in each site to actually 'connect' to the
BridgeHead
Server in the Main-Hub site. (There should actually be TWO
connection objects between them though because connections are
each ONE way.)

There's a connection object on both the spoke site DC(s) and the HUB
sites
DC (which also holds all the FSMO roles)


Now, everythingworks fine, no replication issues at all. My
question
is,
if
the network link broke to the main HUB site, would the KCC figure
this
out
and build new connections to the other sites in order continue
replication.

Well, yes, and no. It would figure out that it had lost connection
to the HQ-HUB site but then it would not be able to connect to
"other sites" (you didn't mention any WANS around or across the
rim) and so nothing would replicate -- this is of course not an
actual issue if you have no other WAN connections.

On the other hand, were the HUB DCs to merely go down (while
the WAN routers continued to work) then the KCC would figure
out that it could use "transitive nature of the site links" to
connect
Rim-Rim sites in pairs -- AS LONG as the WANS and routing
still work.

Sorry, I didn't do a good job of asking that question . I'll try
again
: )
(but thanks for the answers) BTW the betwork is fully routed.

a. If the HUB sites WAN link was down, would the KCC make links
to
the
'other' sites for replication to/from each other?

b. If the HUB sites DC was down/unavailable, would the KCC create
links
to/from other sites for replication?


The automatic feature that makes the site links transitive and
makes this work is (mis)named "Site Link Bridging" (It should
be called "Site Link Grouping" or "Transitive Site Link Groups".

Or does anything else need to be done for this to happen?

How would you expect it to replicate if the WAN is down?

I assumed (wrongly as it will transpire) that the KCC would
dynamicly
build
links between the other sites for replication due to the HUB site
not
being
available. If this is not the case, then what is recommended
practice
in
the event of a WAN failure (albeit they often arent 'down' for
extended
periods)? (When DC's have failed in the past I've manually
triggered
the
KCC on each DC Object in ADS&S)




--
Herb Martin, MCSE, MVP
Accelerated MCSE
http://www.LearnQuick.Com
[phone number on web site]
 
Back
Top