This morning Dell hosted a media breakfast Q&A at the St Regis hotel in San Francisco with Jay Parker (left on the photo), who heads the PowerEdge server team, and Stori Waugh, the group's marketing manager. "We want to renew the dialogue with you all after a period of sticking our head into the sand", said one of the Dell's spokesperson to explain the renew interest by the company to communicate to the press after a long silence since ex-CEO Kevin Rollins departure in January.

The overall theme of the discussion centered around Dell's efforts in server virtualisation and energy efficiency. But in a nutshell: Dell servers are 15% more power efficient than equivalent servers from HP or IBM, virtualisation will drive server sales, blades are no pacea to space constrain data centers and 2U/2 sockets servers or nothing.

Virtualisation will have a positive impact on unit growth, short term, maybe negative after that,

In Q4 the server growth came in slower than anybody had forecast it at the time, much slower. And the conventional wisdom was, virtualisation hits. Our perspective is different. It's not that simple. According to IDC, only 5% of the servers are virtualised, expected to grow to 15% by 2009. It's going to be a long gradual process and could have affect on unit growth over time. But we actually see in the short term some opportunity for unit growth because if you're moving lots of physical servers onto virtual machines, you are not going to do that on a 5 year old server. We see an upgrade cycle.

And here are my other "takeaways" of the meeting:

- Services @ Dell is still a well kept secret!

- Blade servers are not the *only* answer to customers' space/power/cooling issues

HP and IBM are very prescriptive and treating blades as a mandate for customers, not an option. These thinly proprietary systems are designed to prop up their balance sheet as they move customers from RISC units to Itanium or blades. Their business model depends on this lock-in, ours don't. We are committed to blades as an option and you will see us introduce new blade infrastructure chassis systems in the second half of this year that will leapfrog the competition (read IBM and HP!). But that's in addition to our rack and tower servers.

- Dell to continue focus on 2U servers, leave 4U market

70% of our 4 sockets customers could move to our 2 sockets systems with no difference in performance and lower cost. The 2 sockets/2U is really the sweet spot for us. 1U has space limitations like the amount of memory you can have, the number of I/O ports, redundant power, etc. We already choose not to invest in 8 sockets and above platforms. And if you look at market data, they [the 8 sockets systems and above] are pretty much irrelevant at this point.

- Inspite of an easy upgrade path available (e.g. by just switching processors), customers do not touch deployed servers

- Dell will *not* replicate their server lines to support both AMD or Intel chips,

It just does not make sense economically. We'll pick the spot where the technology makes sense or if the product category is big enough to support both lines. Applications that are memory intensive will take advantage of AMD's architecture versus ones that uses more the processor cache will opt for an Intel machine. But often, processors are not the most important criteria in a customer purchase decision, unlike overall power consumption, performance...

- Dell's low flow fan technology was developed by the company's energy efficient lab,

We can regulate the fans speed depending on the server load. When you have 8 fans on a server, that makes a whole lot of difference. The competition can't do that! According to a 3rd party study (I'm looking for the author's name), Dell servers are 15% more power efficient than equivalent servers from HP or IBM