September 2007

I was supposed to be in this session!  I’m so mad…I had another obligation, someone to whom I’d committed to do something, so I couldn’t make the 5:00PM session yesterday during which VMware announced “Enhanced VMotion,” which is designed to alleviate some of the CPU compatibility concerns that have become a bit more visible in the last few months.

In reading the SearchServerVirtualization.com article on this announcement (which is the only reference I’ve seen thus far to the new capabilites), it looks like this may build on top of already-announced enhancements by Intel and AMD (Intel calls theirs FlexMigration, and I think AMD is calling theirs AMD-V Extended or similar).  No specific information was provided on a timeline, but given that ESX 3.5 is currently in beta it sounds like a likely addition to that product’s feature set.

I’ve discussed VMotion compatibility issues here before, so I’m glad to see both hardware vendors as well as VMware addressing these potential concerns.  As I have the opportunity to gather more information, I’ll update this post.  In the meantime, anyone with additional information is asked to add their info in the comments.

Here are some VMworld links that I’ve been collecting over the last day or so.  As I have time, I’ll try to expand upon them and add my own thoughts and views, but wanted to mention them here briefly, at least:

• VMworld 2007: Embedded ESX a can of worms:  Alex Barrett of SearchServerVirtualization.com comments on the release of ESX Server 3i and what impact it may have on the usage of ESX Server in the enterprise.  Certainly it will have an impact, but I think the impact will be greater for third-party vendors than for IT managers.
• VMworld 2007: ESX price is not the problem
• VMworld 2007: VMware users say, “Lower prices”:  So are prices an issue, or not?  The guys and gals at TechTarget seem to be reporting conflicting views.  I’ll have to delve into this in a bit more detail.
• Open Virtual Machine Format Specification (OVF) Submitted to DMTF:  The OVF was announced earlier at VMworld 2007, but no details have really surfaced (at least, none that I’ve seen yet—feel free to forward information you may have found on OVF).
• VMworld 2007: bigger, broader and Viridian:  VMworld, strangely enough, isn’t just about VMware, so here’s some information from Microsoft on their developing hypervisor, code-named “Viridian”.  Here’s another blurb about virtualization security as well.

I’ll try to add some more information and my thoughts on some of these issues as soon as possible.  In the meantime, I’m off to my final session here at VMworld 2007 on Day 2!

I signed up for this session (titled “Advanced ESX 3.0.x Diagnostics Log Analysis”) with the hope of getting some in-depth, useful information on log analysis and troubleshooting.  As a field person, out there in the trenches installing and supporting this stuff, having detailed information on the information that’s being logged by the system is incredibly helpful.  Let’s hope that this session provides exactly that.

The presenter is Mostafa Khalil, a member of VMware Product Support Engineering.  His accent isn’t too bad, which is a pleasant surprise.  He prefaces his session by saying that log analysis is a topic that could span days, but he was only given 60 minutes.  Sounds like a way of saying, “I’m not going to cover everything you want me to cover.”

We start out with a review of the ESX Server boot process, during which process Mostafa indicates that ESX no longer uses the Linux kernel to boot VMkernel.  This is a fairly clear fact now with the release of ESX Server 3i, the embedded version of the hypervisor, but something to note nevertheless.

The easiest way to collect log files is to use the Virtual Infrastructure (VI) Client.  Select an ESX server, go to the Administration menu, and select “Export Diagnostics Data”.  This will collect the following logs (all paths are relative to /var/log):

• vmkernel
• messages
• dmesg
• boot.log
• initrdlogs/*
• vmksummary
• vmware/hostd.log
• vmware/vpx/vpxa.log
• esxcfg-boot.log
• esxcg-firewall.log
• vmware-cim.log
• esxcfg-linuxnet.log
• esxupdate.log (see my patch management session notes for more on this one)
• oldconf/esx*.conf
• rpmpkgs
• vmkernel-version

Questions after the session appeared to indicate that using the vm-support script is equivalent to gathering logs through the VI Client.  If there was anyone else from the session there, can you confirm that’s what you heard as well?

Shortly after this point, my laptop battery died, and so I was unable to transcribe any more information.  I will try to update this article later with some of the information I can remember.  Sorry everyone!  Feel free to chip in for a second laptop battery…

More posts are on the way, so keep reading, and keep the comments and corrections coming!

I’m getting ready to start another session (another super-session) on ESX patch management, presented by Brian Gallob.  Incidentally, I met Brian on Monday night at the VMTN Communities get-together around the corner at the Thirsty Bear (username is BrianG on the forums).  I think I heard there’s about 1200 people registered for this session over in Room 103 of Moscone South, the same room used yesterday for the VC bullet-proofing session I attended.

Brian Gallob started out the session by setting some expectations; he’s not going to make any significant product announcements in this session, and while VMware is working on something to help ease the patching problems, that solution is yet to be announced.  With that expectation in place, Brian proceeded with the session.

Everyone agrees that patching is a necessary evil.  Of course, some users (incorrectly) assume that ESX Server doesn’t require patches, but that’s a false assumption.  ESX Server updates include bug fixes, new drivers, driver enhancements, etc., all of which can be valuable to the end users.

In contrast to earlier patching mechanisms, with ESX Server 3 patches are much more specific, users can “pick and choose” which updates are applicable, and there is (at most) a single reboot required.  Many updates do not even require a reboot.  There is also the new esxupdate tool to help with patching.

Patch notification occurs via e-mail to the e-mail address used during registration of the license; there is also notification posted to the VMTN and the downloads site at VMware (available at http://www.vmware.com/download/vi/vi3_patches.html).  Patches are classified into three categories:

• Security:  These patches fix one or more security vulnerabilities, usually in the Service Console.  They should be installed as quickly as possible.
• Critical:  These fix a bug or flaw in the software.  The flaws that these patches address may cause data loss or severe service disruptions, and should be installed right away.
• General:  Ususally a fix for a minor flaw that might affect a small subset of customers.  These patches should tested and, if applicable to your environment, installed.

Red Hat packages should never be installed on ESX Server.  Although the Service Console is based on Red Hat Enterprise Linux, Red Hat patches won’t work and should be avoided.

The new download site (see the URL earlier) now contains a great deal more information that will make it easier to determine if a particular patch applies to your installation.  The information available there includes:

• Download size
• Brief description of the patch and a patch number (the patch number links back to a KB article about the patch)
• Impact (reboot required?)
• Supersedes or “superseded by” information

All of this information can be used to determine if general patches are applicable.  Security patches and critical patches are recommended to be installed as soon as possible.

With the release of ESX Server 3.0.2, patch installation order is no longer important.  With earlier versions, however, patch should be installed in order by release date.  Within a release date (say, all the patches released in June), order is not important.  The June patches should be installed before the July patches, however.

What’s in an update tarball?  Some RPMs and header/descriptor information used by esxupdate (see below).

Some handy esxupdate commands are:

• esxupdate query:  Shows a list of all installed updates.  Can use a “-l” (that’s a lowercase “L”) to show additional information
• esxupdate info <updatename> - Shows information about a particular installed update; again, use “-l” (lowercase “L”) for additional information
• esxupdate update <updatename> - Installs the specified update; can use the “-noreboot” option to prevent a reboot, useful when installing multiple patches at the same time

For additional information, the /var/log/vmware/esxdupate.log file contains additional information appended to the log every time esxupdate does something.  Tools such as grep (a perennial favorite in these parts, as regular readers will know) can be used to pick up particular pieces of information.  Brian particularly liked these two commands:

grep -i summary esxupdate.log
grep -i totals esxupdate.log

These commands will show more pertinent information on the patches installed, summary information, how many were installed and how many failed, etc.

To use repositories (instead of local update files), simply add the “-r” parameter with the location of the repository, like so:

esxupdate -r file:/path/to/filename update
esxupdate -r http://http server name/path/updatename update
esxupdate -r ftp://ftp server name/path/updatename update

Brian also suggested the use of a shared VMFS3 LUN to store update files.  That’s a pretty handy idea.

Finally, we wrapped up the session with some general troubleshooting tips.  Some useful tidbits from the final portion included:

• don’t use the “-f” option—even if recommended to do so by esxupdate itself—unless you do a lot of pre-work.  If you do, you’ll need to go back and re-apply patches.  This problem is typically caused by installing patches out of order.
• If you get errors trying to untar the update, verify the MD5 sums to ensure the file integrity has not been damaged.  There are Windows-based utilities to calculate the MD5 sum.
• Upgrade to version 2.0.2 of VirtualCenter before you upgrade any ESX hosts to version 3.0.2.  Upgrading to 3.0.2 is recommended as it removes the patch installation order requirements.

Finally, Brian mentioned an upcoming session at 3:30 PM about the future product VMware Update Manager.  I’d love to attend that session, but unfortunately I’m already booked in another session.  I need about 3 more of me around here!

I was so busy yesterday with meetings, VMworld sessions, and other obligations here in San Francisco that I missed out on some important news and posts:

• Thanks to fellow liveblogger Eric Sloof, I learned that VMware has released the majority of the VMware Tools as open source.  A SourceForge site (http://open-vm-tools.sourceforge.net/) has been established for the project.  The official VMware press release is here.
• Mike Laverick of RTFM Education published a great post on VMware Site Recovery Manager, with more details on how VMware SRM works and how it interoperates and integrates with VirtualCenter and the rest of VI3.  Definitely worth a read!  I’m looking forward to more technical details on VMware SRM.  So far there’s been lots of discussion, but I’m a technical guy—I want technical details.
• I had a good conversation with Jeff Woolsey, the Microsoft Windows Virtualization Program Manager, about Windows Server Virtualization and its hypervisor architecture in comparison with VMware’s hypervisor architecture.  I have more research to do yet before I comment more on that particular comparison, but I do appreciate Microsoft (both Jeff and Patrick O’Rourke of Microsoft) taking the time to speak with me and address some of my questions about their architecture.  Both Jeff and Patrick contribute at the Windows Server Division Weblog.

Hopefully, I’ll more coverage today of the various sessions for which I’m scheduled, so stay tuned to the site for more information.  As always, feel free to add your views and thoughts in the comments!

Day 2 is here, and along with it is my liveblog.  Thanks for reading, and I hope that today’s content is useful to you.  Feel free to add your thoughts in the comments below!

Pre-Keynote

I had a great conversation with Mike Laverick of RTFM Education before the keynote started.

Again, about 8:08 AM, the keynote started.  This time, it was no professionally produced video with dramatic natural scenes and inspiring music; no, it was a home-produced amateur video about VMware sung by an as-yet-unannounced gentleman.  I didn’t recognize the song to which he was singing, but the lyrics were actually very well-written.  It was very entertaining.

John Chambers, Chairman and CEO of Cisco Systems

John Chambers starts his keynote out by addressing the pending transition in the markets that are driven by increasingly pervasive connectivity, skyrocketing computing power, and flexible new technologies.  However, these “whiz bang” technologies need to be viewed in the priorities of the CEO and CIO.  Of course, the priorities of the CEO and the CIO may be different, and those need to be balanced and aligned.

<aside>As an aside, I found it interesting that John moved down from the stage and walked among the crowd in the front.  He’s definitely a very comfortable and polished speaker.</aside>

On the importance of catching the transitions—neither being too late nor too early—John discussed Cisco’s use of the Internet back in the late 1990s.  He predicts a “second phase” of innovation, built on virtualization and harnessing the power of the network (Cisco’s network, of course!), creating new services, new support, new sales models, new innovation and acquisition models, and new command/control/collaboration structures.  Embracing this next transition will require the CIO to build an infrastructure, in a cost-effective way, that allows CEOs to “catch the transition”.

One key factor that distinguishes the first wave of innovation driven by the Internet was the direction of information.  In the first phase, information moved from consumers to enterprises and was driven by the enterprises’ decisions; in the next phase, information will flow the other way, and will be driven by the decisions of the consumers.

To effectively harness this, to effectively “catch the transition,” infrastructures must be safe, open, simple, virtualized, and converged.  But all this is just conjecture until John can show us specific, demonstrable products, platforms, or technologies that enable such infrastructures.

It’s interesting to me as John continues his keynote how he uses “virtualization” interchangeably with telepresence (“virtual people”) and remote collaboration.  I suppose that’s a valid use of the term, since you are “virtualizing” your meeting members, but it’s not a traditional use of the term.

Now John begins to bring his vision of virtualization and its connection to network starts become a bit more clear.    It’s about end-to-end virtualization (network, server, storage, resource) that enables a network-orchestrated service oriented architecture.  (Personally, I’m a bit tired of the overuse of the “SOA” term, but that’s just me.)

We finally move away from just talk and into a demo of Cisco’s Virtual Data Center.  John is joined on stage by Jim, the Chief Demonstration Officer (who comes up with these titles?).  In the demo, Cisco shows off a portal that leverages VMware’s VirtualCenter API to automatically provision additional resources.  In conjunction with this, we see Cisco’s VFrame console, which shows us how additional servers running ESX Server were automatically provisioned by VFrame according to a predefined workflow.

Flipping over to VirtualCenter, we see that new ESX Servers and new virtual machines are being provisioned and coming online.  Jim then shows the ANS console, which shows physical utilization and shows how physical resources may be reallocated dynamically as needs change.  Continuing the demo, we return to the Cisco portal and add 400 users to the order processing system, and then we see how VFrame, VirtualCenter, and ANS are going to accommodate the change in demand created as a result.  Like before, additional physical resources are allocated to accommodate the need; and technologies such as VMotion and VMware DRS will distribute the load accordingly.

John wrapped up the keynote with a tie back into business processes and new markets, and how using technologies such as virtualization.  It has to not only be a technology architecture, but a business architecture that empowers and enables the business (and the CEO’s) vision.

Overall, it was a good keynote, but remarkable for the absence of what most people expected from the keynote:  the announcement that Cisco would be shipping third-party switches on ESX Server.  I was really looking forward to this kind of announcement.  So who’s holding up the announcement—Cisco or VMware?

John’s keynote was followed by a panel on energy consumption and energy conservation in the data center.  I’ll blog more on that later on.

The morning of Day 1 of VMworld 2007 was filled with meetings, training sessions, and a self-inflicted mandatory visit to the Solutions Exchange (more on that tomorrow, at which time you’ll understand why I can’t say anything now and why I referred to it as a “self-inflicted mandatory visit”).  In any case, it’s now 2PM on Day 1 here in San Francisco and I just finished standing in line for about 15 minutes to get into a session on protecting VirtualCenter.  Given that VirtualCenter is not a clusterable product that can be protected using Microsoft Cluster Server, anything that I can do to help harden VirtualCenter or increase VirtualCenter’s fault tolerance is a definite plus.

The session is just now getting started, and so far the agenda looks like exactly what I was hoping to see—local availability with VMware HA and MSCS, cross-site availability, DR scenarios, etc.  Excellent!  It looks like the extremely long line for the session will be worth it.

The failure of VirtualCenter will impact VMotion (not available), VMware DRS (not available), but will not impact individual VMs (continue to run), individual ESX hosts (continue to run), or—I was not aware of this—VMware HA.  It turns out that VMware HA, once configured, relies only upon ESX Server for functionality.  In a situation where VirtualCenter is down, VMware HA will continue to operate.  That’s handy.

It is recommended to put the VirtualCenter server, the VMware license server, and the Web Access server on the same server.  This session is going to focus on providing protection for these three components.  The VC database can be protected using “industry standard” methods (read: MSCS or equivalent clustering functionality).

If the license server fails, there is a 14 day grace period, so this means that the failure of the license server component is not a time-sensitive failure that must be addressed immediately.

The VirtualCenter Server is almost stateless; there are only a few files stored locally that must be managed:  the SSL certificate (used to access the VirtualCenter database), the VirtualCenter configuration file, and any upgrade files (these are used to modify the behavior of the VirtualCenter agents on the ESX servers).  There is also the license file, of course, which is used by the license server itself.  Everything else is stored in the VirtualCenter database.

Unfortunately, the second presenter in this session has an incredibly heavy foreign accent, and he’s presenting the information on providing fault tolerance for VirtualCenter.  I hope I can understand what he says.

There are a couple of different ways to use VMware HA to provide protection for VirtualCenter.  Of course, this only works for virtual instances of VirtualCenter, which begs the question of whether VirtualCenter should be run on a physical server or a virtual server.  In any case, using VMware HA to protect VirtualCenter:

• Requires only a single OS instance
• May be subject to network and storage constraints (like what?)
• Will protect you from host hardware failure, but will not protect you against application failure or OS failure

Using MSCS to protect VirtualCenter, on the other hand:

• Requires VirtualCenter 2.0.2 patch 2 or later
• Works for physical or virtual instances (although using MSCS for virtual instances introduces limitations on VMware DRS, VMotion, and VMware HA because of MSCS configuration requirements)
• Requires two OS instances (and requires some file copying and file synchronization, but there should be some easy ways to handle that)
• Needs some additional configuration in order to work correctly

The presenters provided a couple of different ways to use VMware HA, including out of band (using two VC instances in two different VMware HA clusters, each managing the other) and in-band (VC instance on top of a VMware HA cluster managing that same cluster).

When using MSCS to protect VirtualCenter, VMware recommends the use of Majority Node Set (MNS) with a witness share.  No information was given as to why this is the recommendation (I’d love to see that).  Using MNS means it needs three nodes (instead of the two nodes typically seen in an active-passive cluster), but it does enable geographically dispersed clusters.

When it comes to disaster recovery (DR), there are only a couple key points to consider:

• Restoring VC data is just taking care of the database, using whatever methods you would normally use
• Restoring the VC services comes either as a cold standby (reinstalling VC and restoring the state files) or a warm standby (preinstalled but disconnected, with state files being synchronized on a regular basis)

The procedure for performing a failover with a cold standby is as follows:

1. Install a fresh instance of VC on the cold standby server.
2. Install the configuration files and connect to the standby database.
3. Run the Perl scripts to reconnect the ESX hosts, unless the standby server can assume the IP address of the primary VC server (Perl scripts available at http://www.vmware.com/go/viperl/scripts).

With a warm standby, it’s only necessary to ensure that the configration/state files are kept synchronized and then run the Perl scripts (if necessary; see above).

The session wrapped up with some common issues and troubleshooting information, with nothing terribly unusual or outstanding (useful information, yes, but not incredibly groundbreaking).  This included information like necessary network ports, network bandwidth, network latency, etc.  (By the way, you can adjust the network latency timeout by editing the vpxd.cfg file on the VirtualCenter server.)

VMware has announced that it has acquired Dunes Technologies, a company that provides process automation and workflow in virtualized environments (the Dunes announcement is here).  Virtualization.info broke the news about this several days ago, and it was confirmed today with this press release.  Presumably Dunes’ technologies and products will be used to bulk up VMware Site Recovery Manager, Virtual Desktop Manager, and related VMware products.

There’s a great deal of buzz surrounding the VMworld 2007 Day 1 keynote address, in which VMware is expected to make a number of product announcements as well as shed some additional details on product announcements already made, such as ESX Server 3i, VMware Site Recovery Manager, and Virtual Desktop Manager 2.0.  I mentioned all these things in a post yesterday, after completing the first half of Partner Day.  Today, I’ll be trying to liveblog the keynote, so stay tuned for additional details and information as it is presented.

7:50 AM

The conference hall is still filling up.  This year the hall for the keynote looks significantly larger than last year’s venue.  A few other bloggers are already here Eric Sloof of ntpro.nl, Jim Jones of vmwarez.com, Bob Plankers of lonesysadmin.net, among others.  (If I didn’t list you, don’t be offended!)

8:00 AM

We’re still waiting for the keynote to start.  The conference hall filled pretty quickly; right now, it looks like the only spots left are at the far sides of the room, away from any good view of the stage and the numerous projection screens.  While waiting for the keynote to start, I’ve had a great conversation with Jim Jones (of vmwarez.com) about VMware’s new SMB focus.  We both agree that ESX Server 3i is going to make a difference in the SMB market, if only because of the simplicity.

I also just met Viktor, who’s publishing a photo stream of VMworld 2007 via Flickr; go have a look!

About 8:08, the general session started with a heavy duty video presentation and some live music on the stage.  The music was backlit by dramatic natural photos and overlaid with keywords that VMware is driving home—Production Proven, Industry Standard, Better Computing Environment, etc.  Rather nicely done, in my opinion.

Karthik Rau next took the stage, serving as the Master of Ceremonies for the session.  Karthik announced that representatives from both Intel and AMD would be participating in today’s keynote address.  He first introduced Diane Greene, President and CEO of VMware.

Diane Greene, President and CEO, VMware

Diane spoke first about how virtualization is driving a “complete infrastructure refresh”.  She went on to state that the hypervisor is not the same as the operating system, and that virtual infrastructure is not the same as the hypervisor.  I would imagine that these statements are intended to deflect competition from Microsoft and Xen, respectively.

Diane announced again ESX Server 3i, a version of ESX Server that strips away the Service Console, reduces the memory footprint and disk footprint, and allows the hypervisor to run in as little as 32MB.  According to Diane, it also makes VMware’s hypervisor the only hypervisor that does not have any OS dependencies.    (I’d be interested to know how the embedded Xen hypervisor has OS dependencies.  Any Xen experts care to comment?)  Diane then introduced Mark Jarvis from Dell (Dell’s Chief Marketing Officer) to introduce the first ESX Server 3i implementation, the Dell VESO.  They went on to boot, on screen, a Dell server running ESX Server 3i.

Before Mark even finished talking about the hardware, ESX Server 3i had already finished booting.  The Dell appliance was specifically designed around virtualization.  It has more memory slots to accommodate more RAM, double the number of expansion slots, and features the new AMD quad-core “Barcelona” CPUs.

The demo proceeded to use a Remote Console (looked like the VI Client) to find a VM stored on an iSCSI SAN and power it on.  They then went on to show off the CIM standard interface for the hardware vendors to use to interface with the hardware partner’s underlying hardware.  The new Dell VESO system is expected to be available at the end of November.&160; Other vendors that are expected to provide support for ESX Server 3i include IBM (with their X4-based systems), HP, Fujitsu Siemens, and NEC.  There were also a number of video from beta users of the ESX Server 3i product.

Diane discussed the announcement of OVF, an open standard created by VMware after work with Microsoft and Xen and which will be submitted to the DMTF.  VMware believes that the standardization of the containers in which virtual machines are stored will enable a number of new solutions.

VMware Site Recovery Manager is apparently based on OVF and is enabled by OVF to provide workflow and automation.  Diane used Bowdoin and Loyola Marymount as an example of using VMware for disaster recovery, although she stopped short of stating whether these universities were using VMware Site Recovery Manager.

Almost as an aside, Diane also mentioned Virtual Desktop Manager, VMware’s new connection broker that is expected at the end of 2007.

The discussion next moved into virtual appliances, particuarly discussing the LiquidVM appliance from BEA, which strips away everything except those components necessary to run Java applications.  She mentioned again the program from PG&E to rebate customers who implement virtualization.  Two more California-based utilities have already signed on, and 20 more utilities companies are in the process of launching similar programs.  Diane then wrapped up her keynote and introduced Pat Gelsinger, Senior VP for Digital Enterprise Systems at Intel.

Pat Gelsinger, Senior VP for Digital Enterprise Systems, Intel

Pat spent quite a bit of time discussing Intel’s various areas of focus, how those areas map against key IT initiatives and the business needs that those IT initiatives are designed to address.  Some of the specific technologies he mentioned include FlexMigration (the ability to spoof processor IDs to improve live migration compatibility), FlexPriority, EPT (Extended Page Tables), virtual processor IDs, Intel VT for Directed I/O, and Intel VT for Connectivity.  All these are designed to enhance virtualization and improve virtualization’s ability to help satisfy IT initiatives.

Intel VT-d (Intel VT for Directed I/O) provides DMA and interrupt remapping.  This is supposed to be available on all Intel platforms in 2008.

VMDq (Virtual Machine Device queues) is another technology that provides multiple send/receive queues and the ability to pre-sort incoming packets.  This will reduce CPU utilization and improve throughput on network and storage devices.  Intel expects to see, based on prerelease tests, as much as a 100% increase in I/O performance.

Regarding energy efficiency, Intel and Google founded the Climate Savers industry effort to reduce power consumption, and VMware has now joined that effort.  Intel is going to further provide functionality to the VMM (the hypervisor), not only to provide better per-node power management but also to enable datacenter-wide power management.  This would enable scenarios in which we live migrate workloads and power down idle hosts as workload demands decrease.

The Intel Quad-Core Xeon 7300 was demoed (supposed to be included in the HP ProLiant BL680c), and some benchmarks shared about this platform and how it compares to prior Intel CPU platforms.

Hector Ruiz, Chairman and CEO, AMD

Diane next introduced Hector Ruiz of AMD.  He discussed the functionality being added to the Opteron family to ease VMotion compatibility issues across the entire Opteron family.

Moving on to power consumption, Hector discussed the Green Grid, founded by VMware and AMD, and discussed the results of a power consumption study.  In his words, “it’s time to own up to our responsibility to address global climate change.”  By 2010, 70% of every dollar spent on servers will be spent on providing power and cooling for those servers.  AMD itself consolidated their Austin, TX, data center from 117 servers to just 9 servers running ESX Server, resulting in a 79% reduction in power consumption.

The AMD quad-core Opteron processors were officially launched yesterday, and are specifically optimized for virtualization.  Leendert van Doorn, a Senior Fellow with AMD, took the stage along with Hector.

Leendert demonstrated how the new quad-core Opteron currently features nested page tables (dubbed by AMD as “Rapid Virtual Indexing, or RVI) increases performance and helps reduce the size of the VMM.  That, in turn, allows for greater usage of VMMs in more scenarios, on more hardware.  He envisioned the possibilities that could be enabled by having VMMs everywhere, in every computer.  Leendert also envisioned pervasive VMMs as a key technology to enable AMD’s ”50 by 15“ initiative, which seeks to bring Internet access to 50% of the world population by 2015.

Hector’s keynote was heavily focused on both personal and corporate responsibility, true and open partnerships, and continued innovation.  The keynote was short on product announcements, but I personally enjoyed it.

And that’s it!  Karthik Rau took the stage again to recognize other VMworld 2007 Platinum Sponsors and wrap up the general session for Day 1.

I’ll post more information here as soon as I can.  Today is my busiest day the entire week, so updates may be a bit on the light side.

I had the opportunity to attend three Partner Day sessions today, all three from the “Advanced In-Depth Technical Track.”

The first was this morning, and it was focused on an in-depth technical review of VMware Consolidated Backup.  I’ve written a few articles about VMware Consolidated Backup before:

• VMware Consolidated Backup
• Restoring VCB Full Backups with VMware Converter
• Strange VCB Error

The presenter was very knowledgeable; in fact, he was the one running the VCB labs at VMworld (and ran the labs at VMworld last year, which I attended).  I got some great information from the session, and when I have more time I’ll compile that information here.  Some of the key points I took away from the session included information on a command-line interface (CLI) for VMware Converter that allows for automated restores of VCB full VM backups using Converter (I’m really excited about looking into that one); good information on the minimal permissions needed for the user that logs into VirtualCenter and whose login information is hard-coded in config.js (this is a big security concern for many customers); and some RDM compatibility mode issues (RDM in virtual compatibility mode versus physical compatibility mode).  It was a great session; I thoroughly enjoyed it.

Unfortunately, the next session (the first session after lunch) was not nearly as useful.  Although included in the same in-depth track, there was a remarkable lack of technical information.  Fortunately, it was only 45 minutes long.

The final session of the day was on VMware’s VDI and ACE solutions, including a first look at VDM (Virtual Desktop Manager) 2.0, which VMware just publicly announced earlier today.  The presenter, Tommy Walker of VMware, was a great presenter and I enjoyed the presentation.  I’m hoping to be able to catch up with Tommy later this week to conduct some in-depth comparisons of VDM with other brokers, such as Leostream (a broker with which I’ve worked fairly extensively—and which, by the way, just released version 5.0 of their broker).

As soon as I have some additional time, I’ll try to post some additional information about these sessions and some of the in-depth technical details presented.