SQL Server Performance Forum – Threads Archive

Scalable SQL 2000 cluster

I need 2 design a SQL 2000 enterprise solution 2 a customer of mine. We started of with a HP CL380G4 cluster pack (2pcs of HP DL380G4 servers + HP MSA1000 SAN)and the appropriet licenses, 2pcs Windows 2003 enterprise server, 2pcs SQL 2000 enterprise pro. lic. How do i make this solution scalable performance wise.
A classic failover cluster does not boost performance. (active/passive, cant go active/active on a singel database) If one node fails in an share-nothin cluster, i guess the entire cluster goes down? My customer got one singel database @ 100Gb of data. Is partitioning the database over serveral share-nothing nodes the only way to boost performance, do i need SQL 2005 since its a singel database (transparent partitioning) To get failsafe functionality and performance, do i need 2 cluster each share-nothing nodes as classic share node. (each share-nothing node is a classic cluster, means dual hardware per node i guess) Hope someone got a solution for this. many thanks /fredrik, sweden

Clustering is a fault-tolerant solution, not a scalability solution, as you have described. Scability is a big topic and involved many different aspects, including: –Database design
–Application design
–Database partitioning
–Hardware
–etc. How to boost the scability of your client’s database is hard suggest given our limited understanding of it, and how you approach scability will depend on the needs of the client. This website is devoted to covering this topic, and reading through the various tips is a good start in learning about scalability. SQL Server clustering is always a shared-nothing architecture, whether or not it is active/passive or active/active. Shared-nothing refers to the fact that only one node at a time can access the shared array. —————————–
Brad M. McGehee, MVP
Webmaster
SQL-Server-Performance.Com

]]>