Scalability: SQL Server 2005
After you have optimized your database design and the application design, you can improve the scalability of SQL Server in two ways
1.Scaling Up
2.Scaling Out
Scaling Up: Improve performance by
1. Adding more processors, memory, and storage capacity
2. Replacing existing hardware with faster hardware
Note: You also need the right version of SQL Server so that scaling up is supported.
How to scale up Processors:
SQL Server is mostly I/O bound not CPU bound, still better/faster CPU usually provides better performance [depends on the application need though].
Processor choices:32 bit
64 bit x64
64 bit IA64
Hyperthread technology
Multiprocessor based server
Multicore based servers
It is always better to go for 64 bit than 32 bit as it will also support more memory. 64 bit x64 processors provide better performance than ia64 however if your work requires lots of floating point operations and the workload needs > 8 processors you can think about IA64 processors
Multiple processor systems provide better performance than the single processor systems. Windows implements SMP architecture of parallel computing and SQL Server 2005 also takes advantage of it.
Multiple processor systems usually performs better than the equivalent multicore uniprocessor systems.
Intel’s hyperthreading technology may increase performance (10-20%) based on the applications (but not always)
Memory Subsystem:
As SQL server is I/O bound using more memory almost always increases performance. Though the amount of memory that we can add depends on the operating systems. 32 bit operating systems supports 4GB RAM max. For 32 bit windows 2 GB is reserved for operating system, in Windows Server 2003 we can specify 1 gb for operating system so getting 3gb for SQL Server.
32 bit SQL Servers can address 4 gb ram, the actual ram amount depends on the operating system used. if 32 bit windows then 2 to 3 gb max as mentioned before
When 64 bit windows and 64 bit sql server then physical memory limits 32TB.
I/O SubSystems:
You can improve disk I/O subsystem by using an appropriate RAID array or SAN (Storage Area Network) solution
Using Network Interface Card (NIC) teaming, you can also improve performance. NIC teaming allows to bond multiple physical NICs into one logical network device. Teaming provides load balance, fault tolerance, and high bandwidth.
Scaling Out will be discussed later.
Scaling Out involves using multiple SQL servers to provide increased scalability.
From: http://sitestree.com/?p=4804
Categories:38
Tags:
Post Data:2008-04-28 01:08:59
Shop Online: <a href='https://www.ShopForSoul.com/' target='new' rel="noopener">https://www.ShopForSoul.com/</a>
(Big Data, Cloud, Security, Machine Learning): Courses: <a href='http://Training.SitesTree.com' target='new' rel="noopener"> http://Training.SitesTree.com</a>
In Bengali: <a href='http://Bangla.SaLearningSchool.com' target='new' rel="noopener">http://Bangla.SaLearningSchool.com</a>
<a href='http://SitesTree.com' target='new' rel="noopener">http://SitesTree.com</a>
8112223 Canada Inc./JustEtc: <a href='http://JustEtc.net' target='new' rel="noopener">http://JustEtc.net (Software/Web/Mobile/Big-Data/Machine Learning) </a>
Shop Online: <a href='https://www.ShopForSoul.com'> https://www.ShopForSoul.com/</a>
Medium: <a href='https://medium.com/@SayedAhmedCanada' target='new' rel="noopener"> https://medium.com/@SayedAhmedCanada </a>
