That depends on what you mean by touching the tables.
While your snapshot is generated the tables will be read. You can use the
concurrent option which will minimize locking while the snapshot is been
generated.
Other than that the base tables are not modified or read.
There may be some blocking/locking if your server is under heavy load, which
might be what Mike Epprecht is referring to.
In this case you need to use a remote distributor to minimize it.
[quoted text, click to view] "Hassan" <fatima_ja@hotmail.com> wrote in message
news:eyNVrf3vEHA.2596@TK2MSFTNGP10.phx.gbl...
> Since the log reader agent reads off the log file, I am assuming that
> replication never touches any base tables on the publisher directly
> ..right
> ? This is transactional replication and if I did want to use it, my
> frequency would be in continuous mode...Please advice..
>
>
> "Hilary Cotter" <hilary.cotter@gmail.com> wrote in message
> news:%23Oi%23gR2vEHA.3416@TK2MSFTNGP09.phx.gbl...
>> The performance impact is a function of your hardware, the frequency and
>> polling interval of your log reader and distribution agents, the number
>> of
>> transactions and how many rows are affected by each of these
>> transactions.
>>
>> Set up correctly there will be minimal locking/blocking while using
>> replication.
>>
>> "Hassan" <fatima_ja@hotmail.com> wrote in message
>> news:%23cXJipwvEHA.4048@TK2MSFTNGP15.phx.gbl...
>> > We currently use log shipping as a means of reporting for users to
>> > query
>> > off
>> > of on the secondary hence eliminating ad hoc querying on the live OLTP
>> > server
>> >
>> > Problem with log shipping is obvious..Users are kicked off when logs
>> > are
>> > being restored.
>> >
>> > So if we do use replication for each and every table out there and all
>> > OLTP
>> > databases, are there any performance hits on these publishers.. either
>> > from
>> > a CPU/IO/Mem ...We would like to setup transactional replication. Like
> to
>> > know if this is worth implementing or should we just use log shipping
>> > if
>> > there are performance concerns to using replication on the publishers
>> >
>> > Thanks
>> >
>> >
>>
>>
>
>