Server virtualization has had a marked impact on the data center industry, but its associated effects on the network at large may be even greater. A recent Data Center Knowledge report explained that the rise of multi-core CPUs in servers have allowed physical devices to be split into hundreds of virtual machines. Physically, it can be extremely difficult to scale the network to meet the bandwidth challenges created by such dense server setups. As a result, upgrades need to take place from the control plane through the data plane and into the actual hardware infrastructure.
The core problem in virtual server environment is that x86 servers are not designed in such a way that they can be easily configured to perform packet processing, the news source explained. Because of this, secondary hardware systems are needed to provide control plane processing. The capacity to move data packet processing outside of the server and on to specialized network hardware frees the network to accelerate data delivery.
The need to make network upgrades in light of virtualization is especially clear when considering the scale of virtual server infrastructure. To illustrate this, the report detailed a hypothetical facility with 1,000 servers. Within this environment, there are 4 virtual machines attached to each CPU core. With 32 CPU cores in each server, the virtual machine count in every physical device reaches 128, creating 128,000 virtual servers, each functioning as an autonomous machine, in the data center.
Essentially, 1,000 servers equates to 128,000 servers once virtualization is introduced in the hypothetical facility. With an assumed traffic overhead of 1 percent and a 25 percent rate for east-west traffic in the data center, the network management overhead increases by a factor of 32 compared to a traditional, non-virtualized facility, the news source explained.
With added network management challenges existing in the data center, IT leaders also have to address the major increase in bandwidth generated by virtualization. Many experts agree that the process of adjusting to virtualization will lead to the increased implementation of fiber-optic cable systems outside of backhaul and storage networks. As companies consider such a move, they may need to evaluate the potential gains offered by media conversion, which will allow them to use fiber where needed, seamlessly use copper within the broad setup as well, maximizing cost efficiency.
Perle has an extensive range of Managed and Unmanaged Fiber Media Converters
to extended copper-based Ethernet equipment over a fiber optic link, multimode to multimode and multimode to single mode fiber up to 160km.