When Dell designed its 12th generation PowerEdge™ servers, it focused on creating a holistic system. The new line aims to handle increased workloads, optimize virtualization and simplify management, all in one well-designed package. “Balanced system performance is important for a number of reasons,” says Tad Walsh, product line manager for tower servers. “If you put in a new screaming processor with high cycles but you don’t expand the memory capability or IO in proportion, customers will still have bottlenecks.”

Here’s a look at improvements in Dell™ PowerEdge 12th generation servers.

Memory: The Intel ® Sandy Bridge® processor architecture enables higher memory densities than in previous Dell servers. More memory allows for better application performance, even while supporting additional virtual machines and applications per server. A two CPU Sandy Bridge-based machine can support up to 768GB of memory, significantly more than today’s machines. “With the kind of memory footprint available in our new servers, we have essentially eliminated performance bottlenecks,” says Jeff Silva, product line manager for enterprise servers.

Disk IO: The servers feature several technologies to automate and optimize application IO, which can dramatically improve performance for applications. The latest third-generation PCIe technology has higher bandwidth for more data transfer, both outbound and inbound. The servers also support the newest high-performance, flash-based solid-state disk (SSD) technology that connects via the PCIe bus, as well as traditional SSDs that connect through a SAS or SATA interface.

To increase throughput even further, Dell has incorporated its Cachecade™ IO Acceleration into the new PowerEdge Raid Controllers (PERC) for the 12th generation servers. Cachecade enables SSDs to serve as an extension of the cache contained on the storage controller. It automatically moves copies of the most frequently accessed data on a traditional disk drive to the SSD for faster access.

Networking: The new servers offer more flexibility and choice than previous server generations. “There’s a growing migration from 1GbE to 10GbE Ethernet networks, as well as in combining different kinds of networks into a unified network environment,” says Silva. “One of our design goals was to provide as much flexibility as possible for facilitating these trends.”

To this end, Dell is adding more modular networking capabilities with its Select Network Adapter™ (SNA) card. Rather than having fixed network connections soldered to the server motherboard, the SNA is a removable card that connects to a socket on the motherboard. SNAs come in many different configurations and enable users to more easily change the connectivity options on their server as their needs change.

Management: Similarly, Dell is offering more flexibility and choice in terms of itsiDRAC™ (integrated Dell Remote Access Card) management features with a new structure that allows customers to easily try out different versions of the management software or upgrade from one version to another. “We’re combining the functionality of our Lifecycle Controller and iDRAC add-in cards and embedding it all on the motherboards of our 12th generation servers,” says Silva.

By embedding management on the motherboards, Dell is introducing the concept of agent-free system management. With traditional management environments, software agents must be installed and maintained on every server, and an operating system has to be running for those agents to conduct diagnostics. Dell’s embedded management implementation allows administrators to run the same diagnostics on all servers, regardless of OS, and allows for remote management and monitoring.

Many models to choose from

There are six new servers: four rack configurations (the 1U, two-CPU R620; the 2U, two-CPU R720 and R720xd; and the 2U, four-CPU R820); one tower configuration (the two-CPU T620); and one blade configuration (the M620).

Each server brings new capabilities. The R620 server is by definition a denser system, giving SMBs the same memory of a larger system, but in a 1U configuration. “If you use two R620s, you can get twice as much compute capacity and memory in the same rack space, which is helpful for virtualization purposes,” says Paul Steeves, Dell’s product line manager for rack servers.

For the first time, the T620 tower server supports up to four internal GPUs, along with support for two internal PERC controllers. Walsh anticipates these capabilities will support disciplines including business analytics, data information management, CRM, medical imaging and seismic rendering.

--Silicon Valley-based freelancer Howard Baldwin has been writing about business and technology since 1987.