Lower Latency with iWARP
Lowering latency is increasingly important for enterprises because of virtualization, big data applications, Web 2.0 services, and the growing number of real-time applications.
For many of these applications, completing a task is dependent on data calls to servers or storage in other parts of the data center. This east-west traffic is growing, and requires a low-latency infrastructure for the application response time to be acceptable.
For data centers and cloud applications that demand low latency, the network landscape is changing. InfiniBand* is not the only option. iWARP is emerging as a mainstream network technology that can have a dramatic impact on performance and efficiency of many network applications.
iWARP is an attractive alternative to Infiniband, due to the significant investment companies have already made in Ethernet. Any slight advantage InfiniBand might have is diminished when you look at the cost of installing and maintaining a completely separate infrastructure.
iWARP uses your existing Ethernet infrastructure. Now, network managers can install iWARP-capable Ethernet adapter cards in servers, and Microsoft Windows* Server 2012 automatically takes advantage of available RDMA (remote direct memory access) pathways. Because iWARP runs on TCP/IP, no changes are needed to an existing switch and router network infrastructure.
Real-world testing shows that iWARP can match or exceed InfiniBand.
Future Intel® Server Chipsets will support 10 GbE and faster Ethernet with iWARP capabilities, enabling high throughput, and low latency performance at low CPU utilization. Read the white paper for more information about performance.
iWARP vs. InfiniBand* latency