Re: High availability

Yes, I did think of scaling up as an option. I have been dealing with DB2
and Oracle based systems in the past , I recently took over this SQL server
based system, I am trying to understand/explore all options available (since
I am not sure about the options available in SQL server environment).

Currently we are dealing with user environment with couple of hundred users
which is not a problem, the application is running smooth. But in the near
future this system has to be deployed in an environment where couple of
thousand users will be accessing the system (huge jump in volume of data and
number of transactions).

I feel if I scale up, we will be hitting the max performance limits soon due
to sudden increase in number of users. I think scaling out will be more
flexible to add more power in the future if needed. Both High availability
and performance are critical requirements since its an on-line transactional
system.



"Andrew J. Kelly" wrote:

FYI, Replication can be used for non-read only servers but is usually not a
good choice for scaling out in order to get performance gains. You need to
separate the two requirements (Performance & High availability) somewhat as
they are typically two completely different solutions or approaches. What
is wrong with scaling up? How large is the server now and what do you need
to support? And are you sure it has been tuned properly? You might not even
have the need to scale up or out.

--
Andrew J. Kelly SQL MVP

"Suri Nagarajan" <Suri Nagarajan@xxxxxxxxxxxxxxxxxxxxxxxxx> wrote in message
news:BEFF51E6-28E0-4784-A24C-0DBB2C4B497F@xxxxxxxxxxxxxxxx
I am new to SQL server, I would appreciate any help on following question.
I
tried researching and found few options (detailed below after the
question)
which didn't really help what I am trying to do. Any suggestion for the
Gurus
will be helpful.

I am trying to implement a high availability system which will have very
high database access traffic - potentially couple of thousand users trying
to
access a transaction system (read and/or update by most users).
Unfortunately this is not a new system, its an existing system with around
200 tables implemented in a single server (Using SQL server as database),
due to increase in number of users and traffic I need to find ways to
increase availability and performance.

My final system should be implemented on multiple servers and each server
hosting SQL server databases (same data, tables and structures) but they
should be automatically synchronized to have same data (just like multiple
masters in Oracle) so that different users can be connected to different
server (for load balancing and performance improvement . Is this possible
in
SQL server?

I tried following options

1. SQL server clustering - this can only be used for fail over not as
multiple synchronized master databases.

2. Replicated server - replicated server can be used only for read not
for
updates

3. Federated database design with distributed partitioned views - but its
difficult to split certain databases across servers due to complexity and
number of Foreign key constraints defined in the system.


Thanks in advance for suggestions/help.

Regards,
Suri.




.

Relevant Pages

  • Re: High availability
    ... I think you are right regarding scaling up, now I have to convince my client ... But just for my understanding - in SQL server is there an implementation ... like Oracle to have multiple master database in different servers with each ...
    (microsoft.public.sqlserver.clustering)
  • RE: WSS install locked into MSDE DB - Can not install for SQL Serv
    ... For the MASTER database ... Althought we change the Configuration Application pool several times ... interface BUT they appear to be MSDE databases and not SQL Server databases. ...
    (microsoft.public.sharepoint.windowsservices)
  • Re: How to Replicate an SQL Server 2000 Database
    ... actual server name) enterprise manager should associate the database with the ... Looking for a SQL Server replication book? ...
    (microsoft.public.sqlserver.replication)
  • Re: SQL 2005 Best Practice vs SQL 2000: Application Files Separate from data (and log) files
    ... I have had quite a number of clients that used a SAN for a database server and suffered significant performance problems because their underlying SAN infrastructure was a huge stripe across many drives. ... By application files I mean the binn folder which contains the sql server executable among other things. ...
    (microsoft.public.sqlserver.setup)
  • Re: FCB::RemoveAlternateStreams: Operating system error (null)
    ... Thanks for all your help Ekrem, even if your sugestion does not solve the ... in this edition of SQL Server the ... database so it's opened again and you get this error somehow. ...
    (microsoft.public.sqlserver.setup)