Johnny,
when a snapshot is applied, by default the original table is dropped first.
This explains why the table becomes unavailable to subscribers until the
process has finished, and why the FTI is removed. In my reinitializations, I
have a post-snapshot script to add the FTI. If you want continuous access
for the subscribers (and no table-locks on the publisher, transactional
replication might be more suitable.
When the snapshot is applied at the subscriber, this has no relationship to
the transactions required to populate the table on the publisher. The data
is applied at the subscriber in one hit using odbcbcp. I haven't examined
the transaction log at the subscriber using log explorer, but I would expect
to see just one transaction.
Cheers,
Paul Ibison SQL Server MVP, www.replicationanswers.com
(recommended sql server 2000 replication book:
http://www.nwsu.com/0974973602p.html)

Hello,

I have a couple questions regarding Snapshot replication. Ive looked at the
full docs (v. SQL 2005) and understand everything except Im not too clear on
these.

1. While a snapshot of a publication is being applied to a subscriber, are
the replicated tables available in read-only mode with the existing data on
the subscriber? or do they get emptied out first and become unavailable until
the snapshot is applied?

2. Does the subscriber's transaction log grow everytime a new snapshot is
applied? Or does it remain unchanged? For example, if the publisher made
100,000 new transactions to a table. Then it gets replicated to the
subscriber. Wil lthose 100k new transactions appear in the subscriber's
transaction log?

3. When a snapshot is applied to a subscriber table that is full-text
indexed, why does the replication remove the full-text indexing on the table?
Is there a workaround for this to leave the full-text indexing as-is?

Thank you.

[quoted text, click to view]