Hello All,

I have this model in mind. That is presented below.

I am presenting 2 scenarios. Day 1 to day 7, cover scenario 1, and Day
9 covers scenario 2.

Day 1.
A user subscribes and downloads subscription X. The user works -off
line- on his locally cached data

Day 5
The same user decides that he wants to update the local subscription he
has got from day 1. Without having to (check-in) synchronize with the
server. Just pure update of the locally cached (data) subscription. You
can assume for simplicity the delta will cause no collisions with the
locally cached data.

Day 7.
The same user is happy with his work, so he synchronizes (merges) back
with the server. Assume all collisions (if any or no collisions at all)
are resolved.

Day 9.
The same user decides to work again. So he re-initialize with the
server. Can we be selective, say the server - based on history-
figures that the user has pretty much 90% of all the data he needs and
instead of downloading a fresh version, it downloads a delta instead.

You may ask why is that?
Ok, What we have is a huge replica, say in 100,000+++ records. And our
users can be connecting using modems. We don't want to waste time
downloading that amount when all changes that occur were the addition
of 1000 records to the database. So basically, if I downloaded on Day 1
100,000 record and on day 9, the replica would have included 100,1000
records - assume no collision for simplicity-. All I want is those 1000
records added to my 100,000 that are locally cached from Day 1.

Q1:
Is this model a total variant from the normal Merge Replica model?

Q2:
Is this model possible - built-in or by programming functionality it
does not really matter.

Q3:
Can one download a delta (new records added to the database) to the
local copy of the cached subscription from the server?

Q4:
Can we have a subscription that does not expire after synchronization
(merging) with the server?

Q5:
What kind of data transfers can I utilize to expedite the replica
transfer? My data is all ASCII. Can we compress the replica?

Q6:
Can we encrypt the replica after compression?

Please be techinically specific if at all possible.

Thank you sincerely so very much! .

Hello All,

I have this model in mind. That is presented below.

I am presenting 2 scenarios. Day 1 to day 7, cover scenario 1, and Day 9
covers scenario 2.

Day 1.  
A user subscribes and downloads subscription X. The user works -off line- on
his locally cached data

Day 5
The same user decides that he wants to update the local subscription he has
got from day 1. Without having to (check-in) synchronize with the server.
Just pure update of the locally cached (data) subscription.  You can assume
for simplicity the delta will cause no collisions with the locally cached
data.  

Day 7.
The same user is happy with his work, so he synchronizes (merges) back with
the server. Assume all collisions (if any or no collisions at all) are
resolved.

Day 9.
The same user decides to work again. So he re-initialize with the server.
Can we be selective, say the server – based on history- figures that the user
has pretty much 90% of all the data he needs and instead of downloading a
fresh version, it downloads a delta instead.

You may ask why is that?
Ok, What we have is a huge replica, say in 100,000+++ records. And our users
can be connecting using modems. We don’t want to waste time downloading that
amount when all changes that occur were the addition of 1000 records to the
database. So basically, if I downloaded on Day 1 100,000 record and on day 9,
the replica would have included 100,1000 records - assume no collision for
simplicity-. All I want is those 1000 records added to my 100,000 that are
locally cached from Day 1.

Q1:
Is this model a total variant from the normal Merge Replica model?

Q2:
Is this model possible – built-in or by programming functionality it does
not really matter.  

Q3:
Can one download a delta (new records added to the database) to the local
copy of the cached subscription from the server?

Q4:
Can we have a subscription that does not expire after synchronization
(merging) with the server?

Q5:
What kind of data transfers can I utilize to expedite the replica transfer?
My data is all ASCII. Can we compress the replica?

Q6: Can we encrypt the replica after compression?

Thank you sincerely so very much.
MSA

I'm confused. He goes offline, makes changes, syncs on day 7, on day 9 he
syncs, and you want him to just receive the changes which occurred between
day 7 and day 9? This is precisely what happens. You can filter so he only
gets the data he needs.

1) this is the plain vanilla merge model - unless I am missing something.
2) you don't need to do any programming for this, just configure the
subscription
3) sure. Perhaps you might want to configure the ExchangeType 2 (which is
Download Only) to only the transactions which occurred on the publisher flow
to the subscriber. Right click on the merge agent, select agent properties,
Steps, Run Agent, Edit, and at the end of the command section hit space, and
then type -ExchangeType 2
Then accept all defaults and restart your agent.
4) No.
5) use the concurrent snapshot option
6) No, use the encrypted file system for this.
--
Hilary Cotter
Looking for a SQL Server replication book?
http://www.nwsu.com/0974973602.html

Looking for a FAQ on Indexing Services/SQL FTS
http://www.indexserverfaq.com

news:1116348862.558557.136770@g14g2000cwa.googlegroups.com...

1- Let me ask the same question (Q1) with better wording
Can i download a replica on day 1 and then on day 7 download the delta
to that replica?
Delta is the new records added to the server (from day 1 to day 7)
after i dowload the replica. On day I download 1000 record. From d1 to
d7, the server recieved 10 new records. on day 7, i want to download
those 10 record only and add them to my replica. you cam assume no
collision.
Thank you

I'm still confused. Perhaps I am too easily confused.

I think this is what you want.

Day 1, download replica, go offline

In the meantime 10 records are added to the publisher.

Day 7  - sync.

With the exchange type of download only, those 10 records which were added
to the publisher will move from the publisher to the subscriber. Any changes
which occurred on the subscriber will remain there.

With an exchange type of upload only, those 10 records which were changed on
the publisher will remain on the publisher; any changes originating on the
subscriber will move to the publisher.

--
Hilary Cotter
Looking for a SQL Server replication book?
http://www.nwsu.com/0974973602.html

Looking for a FAQ on Indexing Services/SQL FTS
http://www.indexserverfaq.com

news:1116601584.882308.67430@f14g2000cwb.googlegroups.com...