Year: 2007

One of the biggest limit (or benefit, depending on point of views) of VMware ESX Server is its limited support for hardware and software.

For example customers cannot try the product on cheap IDE/SATA lab implementations, cannot replace embedded iSCSI initiator (currently a very old Cisco implementation), etc.

Company policy about HW/SW support is now slightly changing, with the publishing of a community driven Hardware/Software Compatibility List: VMware customers inclined to experimentation can now report which products are working with which version of ESX Server, while other can provide a rating on suggested implementation.

VMware support will act accordingly:

VMware Global Support Services (GSS) will assist customers in problem analysis to determine whether or not the technical issue is related to the 3rd party hardware or software. In order to isolate the Error, we reserve the right to request that the 3rd party hardware or software be removed. This will only be done where we have reason to believe the issue is related to the 3rd party hardware or software.

If VMware GSS cannot directly identify the root cause or it is reasonably suspected that the problem is related to the 3rd party hardware or software, we will direct the customer to open a support request with the 3rd party vendor’s support organization.

It’s reasonable to think VMware will use this initiative to lower testing costs, understanding which 3rd party products are more requested, and using community efforts to recognize which integrations are more reliable, so to move them, after enough QA, to the official ESX Server compatibility guides.

Dell published an interesting 16-pages paper where a 2-way Dell PowerEdge 2950 is compared against the new 4-way HP ProLiant DL 585 G2, both running several virtual machines over a VMware Infrastructure 3 platform:

There is a lot of debate these days around what is the optimal hosting platform for a virtualization deployment. Most of this debate is centered around the decision to deploy either 2 socket or 4 socket building blocks as the basis of the infrastructure. In order to illustrate the advantages of using two-socket servers for virtualization over four-socket servers, a test was conducted with VMware Infrastructure 3 on the Dell PowerEdge 2950 and the HP ProLiant DL585 G2.

The results of these tests show that three PowerEdge 2950 two-socket servers can provide up to 44% more performance, 57% more performance per watt and a 95% average advantage in price / performance than two HP ProLiant DL585 G2 four-socket servers.

Read the whole document at source.

Measurements focus on performance obtained by several Microsoft SQL Server 2005 and MySQL 5 virtual machines running concurrently, and it’s worth to read.

Matthew Brasher, Lead Territory Sales Manager for Enterprise Solutions at Citrix, published on his personal blog a wonderfully funny, unofficial video of some Ardence 4.1 provisioning capabilities, while streaming 3 different operating systems on 150 desktops.

Absolutely worth to see.

Ardence has been acquired by Citrix at the end of 2006.

Parallels continues to improve its virtualization Desktop for Apple Mac OS, now reaching Release Candidate build 3120, and collecting successes, winning the Best of MacWorld 2007 award.

ArsTechnica reported some Apple customers preferences on Parallels over VMware, and several worldwide news sites are write VMware has limited impact on ones interested in Mac OS virtualization capabilities.

ArsTechica also published a nice interview with Ben Rudolph, Marketing Manager at Parallels, revealing some highly wanted new features in a coming major build of Desktop:

…hardware graphics acceleration is expected to come in the next major beta, which as I said in the paragraph above, will be available within the “next couple of months.” Yay! Ben said that other things we can expect are multi-core virtual machines/virtual SMP, some SCSI support, and “a more Mac-like feel.”
…
The next beta will carry even better USB 2.0 support, and there will even be iSight support within Parallels “within the next couple of weeks.” You don’t even have to wait for the next major beta for that one.

Ben told me that the next major beta will also have a more sophisticated Coherence mode…

Read the whole interview at source.

Peter Ferrie, Senior Principal Researcher at Symantec Advanced Threat Research, published an interesting 13-pages whitepaper about detection techniques malware can apply against quite all current virtualization platforms:

…
As more security researchers come to rely on virtual machine emulators, malicious code samples have appeared that are intentionally sensitive to the presence of virtual machine emulators. Those samples alter their behavior (including refusing to run) if a virtual machine emulator is detected. This makes analysis more complicated, and possibly highly misleading. Some descriptions and samples of how virtual machine emulators are detected are presented in this paper.

A harsher attack that malicious code can perform against a virtual machine emulator is the denial-of-service, specifically by causing the virtual machine emulator to exit. Some descriptions and samples of how that is done are presented in this paper…

Read the whole paper at source.

In November 2006 SANS Institute already reported about a range of malicious programs recognizing virtual machines and refusing to run inside them.

A more focused discussion about virtualization insecurities is around the popular Blue Pill proof of concept.
In an interview with Anthony Liguori, Software Engineer at IBM Linux Technology Center, virtualization.info provided a different point of view about this question.

Thanks to GridVM for the news.

Ben Armstrong, Program Manager of Virtual Machine Team at Microsoft, published a useful script to verify which features the virtual CPU assigned to a Virtual Server 2005 has:

Option Explicit
dim vs, vm
‘ Attempt to connect to Virtual Server
Set vs = CreateObject(“VirtualServer.Application”)
‘Get virtual machine object
set vm = vs.FindVirtualMachine(“A virtual machine”)
‘Display virtual machine processor information
wscript.echo “HasMMX : ” & vm.HasMMX
wscript.echo “HasSSE : ” & vm.HasSSE
wscript.echo “HasSSE2 : ” & vm.HasSSE2
wscript.echo “Has3DNow : ” & vm.Has3DNow
wscript.echo “ProcessorSpeed : ” & vm.ProcessorSpeed

Be sure to read original post for updates and comments.

vizioncore published a short but good whitepaper aimed to help virtualization professional in migrating from VMware ESX Server 2.x to the new VMware Infrastructure 3, highlighting 10 major steps:

1. Read all documentation thoroughly before you begin
2. Identify each virtual machine and its associated host for migration
3. Map interdependencies of virtual machines
4. Create a schedule and workflow of virtual machines in required order
5. Identify the method of migration
6. Back-up your existing virtual machines and establish a fail-back strategy
7. Always begin with a test virtual machine
8. Perform migration steps in sequential order
9. Manage your downtime window effectively
10. Leverage automation whenever possible

While some points are really given for assured, others are good advices worth to read. Download it here.

Quoting from the XenSource official announcement:

XenSource, Inc., the leader in infrastructure virtualization solutions based on the open source Xen™ hypervisor, today announced that their virtualization products have received IBM ServerProven certification.
…
Solutions that are ServerProven represent applications that have been enabled for IBM Systems and have been verified through a customer experience to be installed and running in a customer location. To earn the IBM ServerProven certification, XenSource verified XenEnterprise can run on one or more IBM Systems in a real-world customer environment. IBM ServerProven certification is intended for IBM customers to be able to easily identify compatible, reliable solutions for their business-critical needs. XenSource products deliver bare-metal performance on a broad-range of IBM servers, including IBM BladeCenter and IBM System x, running on Windows or Linux.

Virtual Iron was in the process to obtain same status in June 2006.

After VMware (working on VMmark), IBM and Intel (working on vConsolidate) and SPEC (working on a standard), also the popular Center for Internet Security (CIS) is developing a benchmarking platform for virtualization.

But while others focus on perfomances, CIS aims at security measurements, like it already does with several operating systems (from Windows to AIX), network devices (like Cisco routers) and back-end servers (from Oracle Database to Microsoft Exchange Server).

CIS said these benchmarks will be focused on most common platforms, but didn’t provide further details at the moment.

CIS works through a consensus process, which is open to everyone.
Who is interested in partecipating can write directly to John Banghart, Director of Benchmark Services.

J.P.Morgan doubts about real benefits of current relationship between EMC and its subsidiary VMware, as ZDNet reports:

EMC’s acquisition of virtualization software maker VMware completed in January 2004 may have been one of the better deals in the technology sector in the last three years. But perhaps it’s time for EMC to bid adieu.

…

VMware could eventually become a substantial component of EMC’s overall market capitalization. While this suggests that VMware should generate incremental value for EMC’s shareholders, EMC’s stock has not seemed to benefit from this at all.

Indeed, looking at EMC’s market capitalization from 2004 to the third quarter of 2006, we can see that our estimated value for VMware has increased from 2% of the total to 13% in 2006, yet EMC’s stock price fell 10% over the same time period…

Read the whole article at source.

Just few days ago EMC Vice President of Technology Alliances, Chuck Hollis, tried to clarify from his corporate blog why EMC and VMware relationship is so evanescent.