Re: <None>

From: <Tony_Caddies_at_qsp.co.uk>
Date: 15 Mar 1995 13:06:38 GMT
Message-ID: <3k6oov$kb5_at_mailhost.qsp.co.uk>


In <1995Mar13.112947.1_at_orion.yorku.ca>, as360114_at_orion.yorku.ca writes:
>Lately, I have been hearing about "replication servers" related to database
>engines.
>
>With no experience about this item, I am asking for your help:
>
>(1) What is a replication server?
>(2) What applications need a replication server?
>(3) Why is Sybase better than Oracle re: replication servers? Or is it?!
>
>Thanks in advance for your feedback.
>
>Ramy Taraboulsi

Funny you should ask this one. I've just come back from a seminar about replication server and very good it looks too. It could take hours to go into the details of what replication server does to I will restrict myself to answering your questions as briefly as I can.

  1. A replication server is a program which allows you to maintain a fairly up to date copy of data from one database in another. Best illustrated by an example:
	You are running a power station in Location A and you keep all of
	your data in a Sybase database at the location.  However, you
	choose to replicate the data at Location B.  Therefore you run a
	replication server at Location A which passes any changes made to
	it's database to the rep. server at Location B.  If your machine at
	Location A goes down you can quickly switch to your machine at
	Location B and carry on.

   I say that the data is fairly up to date because rep. server works    (basically) by reading the transaction log of one database and passing    the information to another rep. server which applies the same changes    to the other datbase. Obviously there is some lag so that transactions    written by the first database to its logs at the time the machine goes    down may not be transferred to the second database. How big this lag is    depends on how far apart your machines are and what the network between    them is like.

   The example is a *very* simplified one and you can choose to replicate    only certain tables, certain columns in a table, certain rows in a table    or any combination of the above. Another good example is a machine at    a corporate HQ which contains data about all of the regional offices but    only data specific to each regional office is replicated there.

   It *really* can get a lot more complex than this so your best to ask your    Sybase rep. (representative) for some product literature. Sybase have a    very good white paper called SYBASE Replication Server by Alex Moissis    which should answer all of your questions.

2) You need to decide based on your own circumstances.

3) As far as I know Oracle do not have an equivalent replication product,

   but bear the following in mind:

        Rep. server will replicate from Sybase to just about anything:

                e.g. Sybase, Oracle, Ingres, DB/2, Informix etc.

	Rep. server will currently replicate from Sybase but replication
	from DB/2 is in beta test *now* and I understand (don't quote me)
	that replication from Oracle is next with others to follow.


Hope this helps,

Tony Caddies
Sybase DBA
Quality Software Products

All opinions expressed are my own and not necessarily those of my employer Received on Wed Mar 15 1995 - 14:06:38 CET

Original text of this message