Please start any new threads on our new
site at https://forums.sqlteam.com. We've got lots of great SQL Server
experts to answer whatever question you can come up with.
| Author |
Topic |
|
devisetti
Starting Member
30 Posts |
 Posted - 2008-09-28 : 10:47:13
|
| Hi I am designing a database with 11 tables with millions of records.Average records will be 20 million.This database is a readonly database.No DML operations would be performed on this .It is a static data.Can you please suggest me on the following points.1. Is it a good way to partion this type of data.2 If we are not creating primarykey and foreign key relationships.When I write a join between two tables what will be the impact.3.This data base will be accessed by 300 users from a .net application.For search results Only.What will be the best ways to consider the design this type of databaseWill sqlserver 2005 will be best option for this.Thanks in advance |
|
|
LoztInSpace
Aged Yak Warrior
940 Posts |
Posted - 2008-09-28 : 11:07:13
|
| 1) Probably not2) You should define relationships - it gives the optimizer information with which to make efficient queries.3) Great!Your design depends on your application. There is no answer to this 'question' without more detail. However 11 tables is a very small database and 300 users is a very small user base so you would struggle to stuff this up if you follow very basic database practices as documented almost anywhere (3rd normal form, proper relationships defined, proper security, proper data model etc) |
 |
|
|
devisetti
Starting Member
30 Posts |
Posted - 2008-09-28 : 11:36:41
|
| My Worry is about performance each table is having 20 million records on an average.And the front end will be accessed by 300 users. |
|
|
|
LoztInSpace
Aged Yak Warrior
940 Posts |
Posted - 2008-09-28 : 11:59:55
|
| Ok!(You do realise you've given us nothing to go on here?).11 tables - just try it and profile your queries. It's not that hard! |
|
|
|
sodeep
Master Smack Fu Yak Hacker
7174 Posts |
Posted - 2008-09-28 : 13:56:05
|
| If DML operations are not Performed, then heavy indexed will support for OLAP environment for performance.Partition will also play big part in performance. |
|
|
|
saurabhsrivastava
Posting Yak Master
216 Posts |
Posted - 2008-09-28 : 17:18:22
|
| Choose right hardware for this job. Imagine yourself in a situation where everything is optimzed but Hardware cannot scale to processing requirements. Start your database design with SQL Server 2005 because it has lot of new features to help you in this scenario. And keep in mind Partitioning is a manageability feature, not performance. You get performance out of it. Access patterns decides performnce benefits from partitioning. Other considerations for performance improvements; |
|
|
|
saurabhsrivastava
Posting Yak Master
216 Posts |
Posted - 2008-09-28 : 17:21:52
|
| Choose right hardware for this job. Imagine yourself in a situation where everything is optimzed but Hardware cannot scale to processing requirements. Start your database design with SQL Server 2005 because it has lot of new features to help you in this scenario. And keep in mind Partitioning is a manageability feature, not performance. You get performance out of it. Access patterns decides performnce benefits from partitioning. Other considerations for performance improvements; 1) Tempdb optimization2) Number fo filegroups and data files3) Index and data in seperate filegroup4) Storage (SAN or DAS) and RAID type5) Hardware 32-bit or x-64 or IA-64 6) Proper indexing strategy |
|
|
|
|
|
|
Database Design