replication differences between 2005 and 2008

Question

replication differences between 2005 and 2008

foxjazzG

SSCrazy

Points: 2221
More actions
January 14, 2009 at 9:14 am

#203967

What are the differences between these two versions of replication?
Also, if I create a subscriber snapshot database, and script it to subscribe to a publisher, when I go offline, will replication pick up after I come back online automatically, or do I need to jump start it?
Is it better to jumpstart replication (on demand) or have it replicate when connections are "good" to do so?

Viewing 5 posts - 1 through 4 (of 4 total)

You must be logged in to reply to this topic. Login to reply

noeld SSC Guru Points: 96590 More actions · Answer 1

foxjazz (1/14/2009)
What are the differences between these two versions of replication?
Also, if I create a subscriber snapshot database, and script it to subscribe to a publisher, when I go offline, will replication pick up after I come back online automatically, or do I need to jump start it?
Is it better to jumpstart replication (on demand) or have it replicate when connections are "good" to do so?

1. You need to be more specific in your terms what do you mean by "...create a subscriber snapshot database".

2. If what you meant by going "offline" is that you disconnect from the publisher then all transactions (at least on transactional replication) will accumulate on the distributor database for a "retention period" which incidentally is configurable. And when you connect your subscriber backup it will pickup where it left off AS LONG AS it's been less that the maximum retention period otherwise the subscription is invalidated and you will need to reinitialize it (again I am guessing that's what you meant by "jumpstart" it)

Hope it helps

* Noel

foxjazzG SSCrazy Points: 2221 More actions · Answer 2

noeld (1/14/2009)
foxjazz (1/14/2009)
What are the differences between these two versions of replication?
Also, if I create a subscriber snapshot database, and script it to subscribe to a publisher, when I go offline, will replication pick up after I come back online automatically, or do I need to jump start it?
Is it better to jumpstart replication (on demand) or have it replicate when connections are "good" to do so?
1. You need to be more specific in your terms what do you mean by "...create a subscriber snapshot database".
2. If what you meant by going "offline" is that you disconnect from the publisher then all transactions (at least on transactional replication) will accumulate on the distributor database for a "retention period" which incidentally is configurable. And when you connect your subscriber backup it will pickup where it left off AS LONG AS it's been less that the maximum retention period otherwise the subscription is invalidated and you will need to reinitialize it (again I am guessing that's what you meant by "jumpstart" it)
Hope it helps

Yes that helps a lot. Is it possible that we can make a retention period 6 months?

foxjazzG SSCrazy Points: 2221 More actions · Answer 3

foxjazz (1/14/2009)
What are the differences invalidated and you will need to reinitialize it (again I am guessing that's what you meant by "jumpstart" it)
Hope it helps

Is there a method while online with the publisher to ask the subscriber whether or not everything has been pulled/pushed over so that a "user" may validate that replication up to a certain time has occured?

noeld SSC Guru Points: 96590 More actions · Answer 4

Yes that helps a lot. Is it possible that we can make a retention period 6 months?

Well I have never configured to have such long retention period but i theory is possible.

Just make sure that :

@max_distretention parameter of sp_adddistributiondb (Transact-SQL) is that long

AND

that:

@retention parameter of sp_addpublication (Transact-SQL) is also that long or make it "0".

Now, That comes associated with the fact that depending on how many changes you perform your distribution database

can grow to a *very* large size

Is there a method while online with the publisher to ask the subscriber whether or not everything has been pulled/pushed over so that a "user" may validate that replication up to a certain time has occured?

You can check how many "transactions" are still pending to be distributed but no point-in-time resolution is by default verified. You could write some queries based on the latest "transaction datetime" of the records still alive in the distribution database but it just be a ballpark figure, nothing exact!

* Noel