In a recent InfoTECH feature article, I discuss the negative impacts of slow application performance on businesses and how by removing I/O bottlenecks with DRAM storage they can see positive results in employee productivity, customer satisfaction and profitability. Read entire article.
Archive for November, 2010
Posted November 30th, 2010
 |
Boosting Business Applications by Removing Storage I/O Bottlenecks with DRAM Storage TechnologyBy Eyal Markovich |
Posted November 22nd, 2010
|
Measuring I/O Wait – Part 2: SQL ServerBy Eyal Markovich |
In the post “Measuring I/O Wait – Part 1: Oracle”, I discussed how to measure and calculate I/O waits in Oracle. In this post, I will demonstrate how to perform the same calculation with SQL Server. Read the rest of this entry »
Posted November 3rd, 2010
|
Storage Performance and ParallelismBy Eyal Markovich |
Query parallelism processing can push system resources to their limit. Parallelism has its cost (exchange operations, increase in system resources, etc.), and it is not always beneficial to use all available CPUs. It is therefore a common practice to control the maximum degree of parallelism. The optimal maximum degree of parallelism in an OLTP environment will be different in most cases from that in an analytics environment. Obviously, there are many variables in determining the best degree of parallelism, such as the number of cores, server memory, number of available threads, and so on. In this post, I want to emphasize that storage performance is one of the major considerations when setting the maximum degree of parallelism. Read the rest of this entry »
