https://slotsdad.com/ - casino online slots

You Probably Didn’t Notice

by Administrator on September 7, 2009

I really had no use for VM World in San Francisco last week, but after flying six or seven hours to get from Tampa to San Fran, and in anticipation of a ten hour flight back, I figured I would poke my head in on the trade show floor at least.

While the floor was not the size I would have expected from all the hype, there were quite a few smallish booths arranged around three towering anchors.  The latter sported the logos of EMC, Cisco Systems and VMware.  Predictable. 

Most of the booth folk I talked to made it pretty clear that they were just there to “draft off” the marketing money being spent by the big three to promote their mainframe mini-me stovepipes.  NASCAR marketing, if you will. 

drafting

A guy from Brocade said that the show was “pretty successful” in that he had gotten 1500 leads from a small footprint booth.

What you might have overlooked was Xsigo Systems.  Their booth was quite small, but their contribution to the event was huge.  In fact, everything that was going on at the show was courtesy of their Virtual I/O technology. 

Here’s what I mean…

Starting with the VMware booth, Xsigo supported over 50 separate vSphere demos from a single rack of gear located in the center of VMware’s monolithic booth. The ads around the booth hyped performance:  “Capable of running nearly a thousand VMs in just 60” of height, the rack employs dense Supermicro servers, and EMC storage.”  However, central to the rack demonstration and hardly mentioned: Xsigo Virtual I/O.

Xsigo Director (bottom left) consolidates storage and LAN I/O for the VMware Demo Booth

Xsigo Director (bottom left) consolidates storage and LAN I/O for the VMware Demo Booth

Another view.

Another view.

When you consider that last year VMware used 14 half-high racks to power their demos, this represents an incredible savings in cost, space, power and cooling. The difference between the two years is denser server technologies and virtual I/O from Xsigo Systems.  I wonder how happy Cisco was about this — since it demonstrated a much more cost effective alternative to their card-and-switch model in such a dramatic way.  Curiously, no one at the Cisco booth seemed to know what I was talking about when I asked them about being front-ended by little Xsigo.

Even more impressive, to me at least, was the benefits derived from Xsigo in VMware’s demo datacenter.

Central to VMworld's Data Center was Xsigo Systems, which collapsed storage and networking requirements into a much more manageable size.

Central to VMworld's Data Center was Xsigo Systems, which collapsed storage and networking requirements into a much more manageable size.

The VMware Data Center boasted “26 racks of servers generating the I/O equivalent of 5 football fields worth of non-virtualized servers.”  That all of these servers ran on Xsigo Systems virtual I/O, reducing cabling to one or two wires and eliminating the deployment of NICs and HBAs in all servers, went almost unmentioned amazed me.

Xsigo's Virtual I/O simplified cabling, reduced costs, reduced power consumption, and improved I/O performance over what would have been possible with traditional card and switch approaches from Cisco et al.

Xsigo's Virtual I/O simplified cabling, reduced costs, reduced power consumption, and improved I/O performance over what would have been possible with traditional card and switch approaches from Cisco et al.

 

Last but not least, I had a look at performance numbers for Xsigo in conjunction with vSphere and Nehalem processor based servers.  The latter have been touted by Intel as the biggest breakthrough in server chips since Pentium Pro, proffering a supposed 160 percent improvement in speeds and feeds in VMware environments.  Improved server IOPS, plus the purported “breakthrough” speeds and feeds of FC fabrics with FCoE, were supposed to be the stuff of wet dreams for server nerds at the show.  Alas, you wouldn’t have seen much change over your traditional Xeon plus GB Ethernet plus 8Gbps FC connects were it not for little Xsigo…

An interesting chart of Xsigo's speed optimization.

An interesting chart of Xsigo's speed optimization.

A Nehalem server configured with Xsigo I/O, using a 40 gig line rate through a single cable, generated both Fibre Channel and Ethernet traffic. As shown above, the actual data rate of this vSphere + Nehalem server was an impressive 20 gig,  FC + Ethernet traffic combined.

What got me was that NO ONE WAS SAYING ANYTHING ABOUT XSIGO.  Cisco dominated the network news coming out of the show, with its touted ability to failover a VMware server over distance.  (I can do this with CA XOsoft.  So what?)  VMware crowed on about how it was being channel friendlier.  EMC did what EMC always does, hyping its gear’s natural affinity for VMware (big surprise that, given EMC’s 80 percent stake in the company).

The real, and perhaps the only, breakthrough technology demonstration at this show went largely unnoticed.

I wonder why.

By the way, the tchotchke at the show was lousy – pens and product literature – and my kids were disappointed. I just explained that all of the toys at this particular event were “virtual.”

They said that virtual toys are lame.  Out of the mouths of babes…

Previous post:

Next post: