At VMworld 2012 in San Francisco Hiro Matsunami, Senior Manager Product Management Servers & Systems Software Division and Chas Weber, Director Strategic Alliances of NEC briefed me about NEC’s upcoming R320c fault tolerant server. The new R320c is the sixth generation of NEC fault tolerant servers. NEC first began manufacturing fault tolerant servers back in 2000. Unlike general purpose servers which are rendered inoperable when a major component fails, fault tolerant servers like the R320c are designed to withstand all types of component failures with no downtime. The R320c has fully redundant system components. Internally, there are two system modules each of which contains its own motherboard CPU and RAM. The two modules run in lockstep with same instructions executing on both the system modules at the same time. If there is a component failure in one of the system modules the other module will continue running with no interruption of service. The R320c provides 99.999% uptime right out of the box.
The new R320c is a 4U system that features two eight core Intel Sandy Bridge Xeon E5-2670 CPUs. It supports up to 256GB of RAM and four 10GB onboard Ethernet NICs. In addition, the Power Supply Unit (PSU) is compliant with the 80 PLUS Platinum standard.
Hiro and Chas shared that one of the most frequent questions that they received during the VMworld 2012 conference was about the differences between NEC’s R320c fault tolerant server and VMware’s software based Fault Tolerant feature. Although the technologies are similarly named and have similar goals they are implemented very differently and have quite different characteristics. VMware’s Fault Tolerant feature only works at the virtual machine level and it only supports a single virtual CPU making it unsuitable for resource intensive workloads. Multi CPU support is in the works but it is not currently available. Further, because VMware’s Fault Tolerance is implemented in software there is a performance penalty for implementing it. Fault tolerant servers like the NEC R320c have multiple CPU and multi-core support built-in and because the fault tolerance is implemented at the hardware level there is no noticeable overhead or performance penalty. In addition, when the fault tolerance is implemented at the server hardware level you can use it to provide protection for virtualization hosts like vSphere Server or Hyper-V which will also help to protect all of the VMs that run on those hosts.
You can read more about our lab testing of the previous release of the NEC R320B fault tolerant server at: http://www.windowsitpro.com/article/product-review/product-review-nec-5800-fault-tolerant-server