Fast transference of data from one peripheral to another is the requirement of today’s technical environment. Keeping this need in mind, Intel has brought in I/O acceleration technology which assures the fast transfer of data.
 
To assure the improved data transfer rates IT managers keep on spending large funds to attain the required speed. But on network, data transference has yet other bottlenecks to fight with. Intel’s I/O acceleration Technology assures the enhanced rate of moving data on network and receiving the full output of what is required from the server investments.  

I/O acceleration technology can be precisely taken as the technology that accelerates the transfer of data on network. I/O acceleration technology along with the Quick Data technology which is available in latest Quad Core and Dual Core Xeon processor-based servers address the bottlenecks of network for fast data transfer rates.

Contrary to earlier approaches that were more or less the point solutions to fix the absence of enhanced performance, I/O acceleration technology is developed using wider approach. It is a platform based approach that specifically addresses the lack of required speed after investing so heavily on network and servers.

The technology is much appreciated by IT managers attributing to its enhanced performance, ability to address the network I/O bottlenecks, scalability, flexibility, and reliability. The technology reduces the number of CPU overheads, and thus frees the CPU for more critical tasks. In brief, this technology can be termed as a comprehensive solution for more complicated problems.

To understand how this technology helps to achieve faster data rates on network, one has to understand the basic mechanism of how data is transferred on network. For data transfer on network basic TCP/IP protocol suite is applied. In so many years, when the amount of data has reached massive heights, the basic mechanism behind the protocol remains the same. As a result, CPU overheads increases and the available number of CPU cycles reduce for applications, thus degrading the performance and user experience after investing so heavily on appliances.

To address the I/O bottlenecks, Intel’s I/O acceleration technology uses three-pronged approach:
1.    Works in a direction to reduce the system overhead
2.    Streamline the process to access memory
3.    To optimize the process of TCP/IP protocol computation

System Overhead Reduction: I/O acceleration technology reduces system overhead by reducing the number of context switches through interrupt moderation, moderation the memory access procedure, and achieving parallel operations. The data movement procedures and memory access procedures are moderated by implying the pre-fetch mechanism, Direct Memory Access (DMA) and creating sync among the data flow and specific processor cores.

Streamlining Memory Access: The basic methodology behind the I/O acceleration technology is the reduction in number of memory access and number of interrupts thus allowing CPU to compute processes more efficiently. As the technology also make use of Quick Data Technology, the data movement primarily occurs between memory controller and main memory thus freeing a fair amount of FSB bandwidth. This allows CPU to overlap data movement with other operation requiring computations thus making I/O faster.

Optimizing TCP/IP: The technology optimizes TCP/IP computation by reducing the number of access made by CPU to application buffer. It allows an application to send an entire block of data at once to Network Interface Card (NIC) instead of sending a block.

The above discussed three way approach makes I/O acceleration technology a system wide approach thus enabling higher performance, reliability and efficiency.

At last it can be summarized that I/O acceleration technology is a great boon to organizations that require rooting out the network bottlenecks for faster I/O while managing the increased network load. The technology is natively present with Microsoft Windows server 2003.  

Renu Singh/ITVoir Network