In a converged data center environment, traditional I/O and storage connectivity, along with LAN and data communications, are highly interdependent. In general, the faster a processor or server, the more prone it is to performance issues when waiting for slower I/O operations. As a result, faster servers need better-performing I/O connectivity and networks. Better performing means lower latency, more I/O operations per second (IOPS) and better bandwidth. This higher level of performance is required to keep up with various operations and application profiles and to avoid bottlenecks and aggravation caused by aggregating data center resources.
Blade servers support traditional LAN and storage connectivity. But they also are prime candidates to leverage converged network adapters (CNAs) that support both legacy Ethernet TCP/IP- and TCP/UDP-based traffic, along with Fibre Channel over Ethernet (FCoE). This means that a single CNA (or, per best practices, a pair for redundancy and performance) provides the same functionality as a traditional Fibre Channel or LAN adapter, reducing cabling complexity while freeing up valuable expansion space within the servers themselves. A single CNA card can support, concurrently, traditional IP functions over Ethernet as well as iSCSI, NAS and Fibre Channel (via FCoE) in order to meet application-specific needs while at the same time implementing QoS and other functionality. CNAs are also flexible because IT can reconfigure them to meet changing needs as blade servers are reprovisioned from physical to virtual support, or for different application requirements.
Moving forward, converged, enhanced Ethernet, supporting FCoE, provides the ability to carry Fibre Channel traffic — including Fibre Channel Protocol (FCP) and Fibre Connectivity (FICON) — via Ethernet while also transporting TCP/IP-based traffic. This differs from current approaches where Fibre Channel traffic can be mapped onto IP using Fibre Channel over IP (FCIP) for long-distance, remote replication. With FCoE, the TCP/IP layer is removed, along with any associated latency or overhead, but only for local usage.
Bandwidth optimization solutions help systems coexist in hybrid environments, enhancing backup or the movement of distributed data and accessing cloud resources. When moving data to or from cloud-based or remote backup services, bandwidth optimization can take the form of data footprint reduction at the source, as well as protocol and network technologies.
For more tips and information on data center covergence, download our Data Center Convergence White Paper.