The Case for Scale-Out NAS

Date Added: Jul 2009
Format: PDF

It is not surprising that data center bottlenecks cause serious repercussions to a company. This often leads to customer dissatisfaction and loss of important contracts. Bottlenecks increase when more aggregate workloads are incorporated into servers that don't have the sufficient memory to do it. When the CPU's processing power goes down, there is lack of available memory and undersized I/O interfaces are the nightmare of any business enterprise. Seasonal workload surges like e-mails, files, videos, applications, web usage and database all create bottlenecks. These deadlock conditions will result in poor result time that will lead to unreliability when serious workloads come in. The paper prepared by George Crump, Senior Analyst in Storage Switzerland discuses the cause of I/O bottlenecks, what causes it and how it is caused. He addresses the necessity of handling these bottlenecks in critical situations by making use of fast running processors. The speed of the disk system should be increased to handle all the data transfer in real time. Network Attached Storage (NAS) controllers are responsible for data storage in the disk systems and when the demand exceeds the supply ratio, there is serious trouble and the company is headed for a downfall. If servers and networks are used intelligently, then it is equally mandatory have a fast paced NAS controller on those disks. Unless that is rectified and multi-tenant workloads are eased, there is still the issue of bottlenecks.