I use replication a lot
I use replication a lot, but not so much that I wouldn't benefit from a good
book or two. I recently purchased Pro SQL Server 2005 Replication (Definitive
Guide) (Hardcover) by Sujoy Paul and rather than my standard attempt at reviews of books I buy, thought instead I'd try an interview format. Overall I
consider it to be a good book, but still left me wishing for a bit deeper
coverage in some of the advanced areas. Good examples and covers both GUI and
Anyway, on the with the interview!
AW: How long have you been working with replication?
SP: I have been working with replication for almost 11 years.
AW: What made you decide to write a replication book?
SP: As I was working with SQL Server I realized there was a dearth of books on
replication that encompassed the whole gamut of different types of replication
namely configuration, administration and monitoring from a real-life
perspective. I wanted to write a book that will focus not only on the usage of
traditional tools like GUI and T-SQL but also leverage new tools like SQLCMD,
tablediff and DTA in the configuration and administration of replication. So
while the initial chapters will deal with the configuration of different
replication, followed by the internals I also wanted to show, for example, how
to use the power of SQLCMD as in the case of no-synch initialization of merge
subscription or configuration of log shipping for merge publication with
download-only articles. Readers would then be able to make use of such examples
in other scenarios wherever applicable. Based on the numerous emails that I
received, I hope I have been able to achieve that.
AW: How long did the book take to complete?
SP: It took me almost 15 months to complete the whole book.
AW: 15 months is a long time to work on a
project. Do you plan to update the book for SQL 2008, and do you have any other
SP: I need to take a look at SQL
2008 before I can decide accordingly. Yes, I would like to write some more but
haven't made up my mind on the topic as yet.
AW: One of the things I noticed about the book is that each replication topic
you have two chapters; one using the GUI tools and one using TSQL. What made you
take that approach and why do you think it's better than just covering it all in
one chapter (or perhaps leaving out the TSQL portion)?
SP: I have taught database and data modeling for several years. The one thing
that I had noticed over the years was that people would always play with the GUI
before getting into the nitty gritty of coding. To think of GUI as mere
screen-shots is to miss the forest for the trees. The availability of GUI helps
the user to appreciate the beauty of constructing the replication process
without getting into the nuts and bolts of replication. It then becomes easier
and interesting to learn how to code using TSQL. Replication is a vast canvas
and, personally, T-SQL is the brush by which you can paint. Practitioners of
replication will always try to master all the different techniques to marvel the
way it works in different ways.
AW: What is your overall favorite type of replication?
SP: That’s a tough one. I think it also depends on the business criteria.
Non-hierarchical replication would be one of those replication types that you
can also use for other purposes. Merge replication with its associated conflicts
and maintenance is also a challenge that I relish.
AW: What is your favorite new replication feature in SQL 2005?
SP: I think the introduction of non-hierarchical replication like peer-to-peer
would definitely be the one at the top of my list of favorites.
AW: What changes in replication are you looking forward to in SQL 2008?
SP: Quite a few. First and foremost would be the ability to transmit messages
using MAPI and/or SMTP so that data can be send/received via email. This would
particularly be useful for SFA application where sales people can retrieve the
data remotely via email. There are some other nice changes coming as well:
- The ability to change schema and send to subscribers without
regenerating the snapshot
- The ability to configure the number of engines that can be specifically
used for replication
- Web-based monitoring of replication and make replication more
- The ability to design replication where we can have, at least, something
similar to the CRUD matrix used in database design
- The ability to perform heterogeneous replication across all databases
not just Oracle or DB2
AW: On the back of your book you refer to the complexity involved in
replication and the associated jargon. Do you think that complexity leads
users to other solutions such as SSIS that may be perceived as easier to
work with when replication would really be a better solution?
SP: Well, SSIS is a powerful ETL tool that can stand on its own. I don’t
think SSIS and replication are complementary to each other. I think the
simplicity of the execution and the beauty of the process lies in the
complexity and the detail involved. To me the holy grail of bi-directional
data transfer across different locales is replication.
What is the most unusual or complex replication set up you have worked on?
I think replicating across heterogeneous databases. It’s not unusual but can
AW: Will you be at the PASS Summit in Denver (Sep 2007)?
SP: I still would like to but prior family commitments might preclude me
from attending the summit