New Servers Pack in the Power

Arthur Cole

It looks like the march of the bigger, badder server is hitting full stride, as the drive to consolidate as much computing power on as little hardware as possible picks up steam.

A number of key new systems and underlying CPU architectures are on the way out in hopes of tapping into both the financial and environmental pressures that are set to increase consolidation ratios in data centers around the world.

First, we have Sun and Fujitsu, which just came out with an improved version of the Sparc Enterprise M3000 server. The device supports quad-core Spar64 VII processors plus a new generation of memory module that, combined, bump clock speeds from 2.52 GhZ to 2.75 GHz, a 23 percent gain. The 2 RU device runs Solaris 10 and is optimized for any number of enterprise applications, such as databases, CRM and ERP. And even though the companies have not yet listed a price, they claim the M3000 will still qualify as an entry-level machine.

The drive for improved performance and greater scalability is leading some vendors to roll out new systems even as enabling technologies are barely out of the lab. IBM did as much recently when it announced a pair of servers using Intel's Core i3s, which were just then being announced at CES. The x3200 M3 tower supports the new dual-core i3-530 and 540, which clock in at 2.93 and 3.06 GHz respectively. Still, since the machine also supports the Pentium G6950 and Celeron G1101, neither of which have an integrated graphics processor like the i3, M3 boards will sport a Matrox graphics controller with 16 MB of video memory.

But even the new i3s might not represent the state of the art for very long. Intel is said to be planning the launch of the 32 nm Westmere Xeons within the next few months. These are said to boost performance and lower power draws compared to current Xeon technology, and will include integrated graphics and AES (Advanced Encryption Standard) security to help protect data residing in virtual or even cloud environments.

Servers are likely to retain their commodity status as more powerful machines hit the channel, but it is also widely expected that they will come pre-configured for specific workloads or environments, removing much of the integration and setup that dominates deployment scenarios. That means fewer boxes to be managed in the data center, but increased pressure to ensure their reliability.

Subscribe to our Newsletters

Sign up now and get the best business technology insights direct to your inbox.


Add Comment      Leave a comment on this blog post
Jan 22, 2010 9:01 AM Paul Huber Paul Huber  says:

A GE division I worked in consolidated 50 servers to eight servers running sixteen domains, N+1, 3 to 1 failover dedicated failover, 5 to 7 Oracle instances per node, on mid range (SUN OS and Veritas Cluster) in 2002.

Citibank did a global consolidation to 25K for floor space in 2004, where the Enterprise class servers could replace 24 physical servers, 12 with no single point of failure (two system boards) the concept of one server one application has been dead for quite some time, HP has NPARS, IBM PowerVM and LPARS, the concept is called consolidation.

Jan 22, 2010 3:46 PM KLC KLC  says:

It would be interesting to see a performance oriented comparison of virtual hosts, including:

VMWare, XEN, Hyper-V:

+ Windows

+ Linux

UNIXs - AIX, HP, Solaris

Mainframe z/VM - z/Linux, z/OpenSolaris, z/Windows

Anything I missed.


Post a comment





(Maximum characters: 1200). You have 1200 characters left.




Subscribe Daily Edge Newsletters

Sign up now and get the best business technology insights direct to your inbox.

Subscribe Daily Edge Newsletters

Sign up now and get the best business technology insights direct to your inbox.