I’m taking a breather from re-recording the voice track for a Video on Live Migration in Hyper-V. When it’s done it will end up on YouTube.
Now YouTube is giving me pause right now: it is certainly the easiest way to put up videos so that people can find them. I’m viewing it as an experiment because anecdotal evidence suggests that the audience I’m trying to reach (IT Professionals) don’t really think of it as a good source of content. And also because you-tube is all very well for bits of fun (like this one or these ) but if you have a serious message isn’t it a bit needy ?
Consider the case, for example, that you have had things pretty much your own way for years, but now it seems everyone who hasn’t deserted you yet is flirting with the other side, the people who pay your wages are even beginning to question the money they pay you. This may sound like Gordon Brown’s attempts to woo the public – although derided by his colleagues* at least he has some more control over YouTube than press photos - but in fact I was referring to VMware.
There’s a video on You tube which starts with a factual error about Technet and MSDN. First off, Technet and MSDN themselves didn’t fail, the failure was in the download site for Windows 7, Server 2008 R2 and Hyper-V Server R2. Demand for these greatly outstripped predictions – there just wasn’t enough hardware capacity, as explained here. I don’t have the stats on how many of the downloads were for Hyper-V server R2 or people wanting to test Hyper-V on Windows Server 2008 R2, we’d been averaging in excess of 100,000 downloads per month of the first release of Hyper-V. Live migration was missing from that release (the main reason that customers chose VMware) but its in R2 - even with the free Hyper-V server product (hence my video). These must be scary times at VMware… but I digress.
Last year I ripped a hole in VMware’s dishonest pricing examples, this year one someone thought it would be a wheeze to post footage of Hyper-V blue screening on YouTube. He kept quiet about who he was, but it didn’t take long for Virtualization Review to reveal “the root cause of the VMware FUD: Scott Drummonds.” as they go on to say “his job basically is to look at the competition and spread the word about VMware superiority. Unfortunately, Drummonds doesn't identify himself on the Hyper-V crashing video. Why not? Cynics might say because the video would have less impact if they knew it came from Microsoft's chief virtualization competitor”.
Drummonds confesses he made the post and gives some blather about using two virtual disks (in VMware IDE performance isn’t much good, so they run the test from SCSI disks. In Hyper-V IDE performance matches SCSI yet they wanted to run the test from SCSI disks, which conveys a degree of ignorance of hyper-V and a lack of scientific method - what effect does doubling the number of disks have on the validity of the tests ?).
Jeff Woolsey demolishes the Video – since it showed the STOP error at the blue screen he went digging and found from our 750,000 downloads of hyper-v “we've had 3 reports of crashes under stress and with the same error code as seen in the video bugcheck (0x00020001). The solution in all three cases was to upgrade the server BIOS which solved the problem”. VMware have seen similar things incidentally. I love a good demolition, so Jeff’s follow-up post makes good reading; in particular he points out that to have any merit a test has to be repeatable, there’s no published methodology, no statement of what is in the VMs being tested, what the hardware was etc. Jeff points out that VMware prohibit publication of benchmarks unless they have approved the way in which they are carried out, because as they put it Benchmarking is a difficult process fraught with error and complexity at every turn. It's important for those attempting to analyze performance of systems to understand what they're doing to avoid drawing the wrong conclusions or allowing their readers to do so."
As the question of disks made clear they don’t understand they are doing with hyper-v and anyone doing a serious test for publication would put in a call to Microsoft support and get a problem like this solved. Who says “A-ha ! a problem with the competitor let’s not try to fix, just video it and put it on YouTube.” … well Scott Drummonds, obviously. But you can decide for yourself if VMware – at least Scott – were allowing their readers to draw the wrong conclusions or deliberately leading them astray.
Oh and Scott, if you’re reading, anyone with a good knowledge of testing windows will tell you that SlMgr.vbs –rearm will stop that “You may a victim of counterfeiting” message spoiling your videos.
* I had to go to the source of Hazel Blears’ comment “You tube if you want to” because it sounded like there was something missing it’s obviously meant to echo the famous Thatcher quote “To those expecting a u-turn I say You turn if you want to … the lady’s not for turning”. The full quote is actually “I’m not against new media. YouTube if you want to. But it’s not substitute for knocking on doors”.
Update. Fixed some typos and bad edits.
Great post, it's good to see someone pulling VMWare up on their propaganda...my VMWare preferring colleagues are just as bad. This'll definitely help when it comes to mindshare.
The propoganda is happening both ways and neither party (MS or VMware) is innocent. Who cares who started it, just stop as it is boring and childish. Move on already.
Talk about the tech specifics people want to see, the best practises and all the useful information.
Have you read the latest response from Vmware, they clear things up and show their entire testing methods. Your little assumption of their hyper-v ignorance just got 'demolished'
They tested with IDE but used scsi because it gave better performance. They gave hyper-v every advantage in the test, and it still failed.
Shawn, they didn't actually. They tested with two disks, IDE + SCSI instead of one disk IDE.
I've read the post - no comments were allowed, and it's pretty critical of Drummonds.
Did they call Microsoft to get the issues solved or follow any of the things they say need to be done in preparing tests for publication.
They did do IDE only...
Quote from Bruce Herndon's repsonse:
"I think most folks would be interested to know that the in the course of our extensive research we tested just such a scenario. In this case the system was a Dell 2950 with dual Intel Xeon X5460 (3.16 GHz) and 32GB of RAM. As you can see below, IDE-only exhibited higher utilization (except at saturation as expected) and lower throughput at every data point. Based on these and other results, we standardized on the IDE/SCSI dual-disk solution to give Hyper-V every possible advantage.
I thought they tested both scenarios,
"As you can see below, IDE-only exhibited higher utilization (except at saturation as expected) and lower throughput at every data point. Based on these and other results, we standardized on the IDE/SCSI dual-disk solution to give Hyper-V every possible advantage."
What happened to my last comment?
Shawn, sorry I can be a bit slow checking and approving comments, and it's worse round the weekend.
They compared ONE drive against TWO drives. On the basis of comparing a single IDE against an IDE and a SCSI together they said SCSI was faster. Without the comparison point of two IDE drives that's not a valid conclusion. I would never have passed the science exams I did aged 15 with method like that.
Who is to say that if they had run a 1xSCSI + 1xIDE config on VMware that wouldn't have blue screened.
Etc etc etc.
There is also no getting away that VMware is behaving like a 2 bit company, not a billion dollar one. (see Jeff for more on this http://blogs.technet.com/virtualization/archive/2009/05/17/vmware-fud-fiasco-part-3.aspx )
and anyone doing an objective test would have talked to the company who's software they are testing before rushing to you-tube
Sorry. I can't help being amused by Microsoft, the inventor of FUD playing the innocent card.
There will be endless arguments about bios versions, IDE/SCSI and who knows what else. However at the end of the day, this video shows a) a cascade failure within supposedly isolated VMs b) failure of the entire host c) that this was based around performance load i.e. it didn't happen all the time.
I would argue that these things shouldn't be able to happen regardless of whether its IDE/SCSI etc.
We credited with inventing all sorts of things. But to quote wikipedia
FUD was first defined (circa 1975) by Gene Amdahl after he left IBM to found his own company, Amdahl Corp.: "FUD is the fear, uncertainty, and doubt that IBM sales people instill in the minds of potential customers who might be considering Amdahl products." The term has also been attributed to veteran Morgan Stanley computer analyst Ulrich Weil, though it had already been used in other contexts as far back as the 1920s.
If the VMware have a reproducible method to crash Hyper-V they would publish the method. They got an error which has only ever been reported with an out of date bios ; let's not kid ourselves: if Hyper-V were crashable on demand every VMware presales support person would be crashing it in front of customers every day of the week.
VMware shows all the signs of running scared. The thing is, when you do that you need to watch yourself or you fall in the gutter.
And to also quote wikipedia,
"Although once it was usually attributed to IBM, in the 1990s and later the term became most often associated with industry giant Microsoft. Said Roger Irwin:
“ Microsoft soon picked up the art of FUD from IBM, and throughout the 80's used FUD as a primary marketing tool, much as IBM had in the previous decade. They ended up out FUD-ding IBM themselves during the OS2 vs Win3.1 years."
Point is that everyone knows that Microsoft employs FUD (I gave some examples in the previous post). Therefore for them to try and take a righteous position when someone is giving them a taste of their own medicine is somewhat laughable and hypocritical.
Paul, I notice that talks about the 1990s or is it the 1980s. When Paul Maritz (VMware CEO) was held a senior post at Microsoft. If you're saying VMware is behaving like the Microsoft of the 1990s I might have to agree. We came to see that as wrong Microsoft behavior today isn't the same as it was when I joined in 2000. (Although claiming no current Microsoft employee ever resorts to FUD would be difficult). You do seem to be arguing that two wrongs make a right.
A few weeks back I posted about VMware’s conduct in posting a video which of some tests which appear