hotstuff

you do? how does it work? please elaborate, i am skeptical, primarily because i looked into coding this and it is *very* hard to deal with disk io coming from page cache ops and because your company has a track record of bogus technical claims. my mind is open however - prove me wrong.

your management tools are good, i already stated that. at the time i was considering virtuozzo, they cost 1-2k iirc. has this changed?

this is blatantly incorrect. both ram size and cpu qos is changable with one command each. its right there in the xen documentation. did you just not bother to read or are you doing this deliberately? dont you think its a bit low to disseminate completely incorrect information about your competitors' products? if you havent noticed, the adult industry is big on integrity and honesty and you arent doing yourself any favours in either of those departments.

as far as network io, you can use whatever method floats your boat with xen - no support for this from xen is required.

bzzzzzzzzzt, wrong answer. you can migrate xen vms perfectly fine without shared storage. again, its right there in the docs.

yes, youve got nice mgmt tools. novell, ibm and sun are working on theirs for xen. xensource also has something in the works. this is one area you have advantage in.

only partially correct. there is a xen implementation for windows, but it can not be released due to licensing issues. who cares about windows boxes anyway? the only thing they're good for in this vertical is drm.

again, mostly incorrect. whether you can save on licensing costs depends on the license, and most licenses do not in fact allow you to save money in this scenario - your costs might be higher.

it is perfectly possible to achieve the same (or even better) hdd space utilization in xen using lvm copy on write, without the dubious benefit of your filter filesystem. id rather use something i can trust, like ext3, than to keep my fingers crossed that your coders didnt fuck up interaction with vfs - filter filesystems are not trivial.

virtuozzo uses *the same kernel* for all virtual environments (which is a problem, as i will explain later). this allows your product to save ram on some application pages. very small benefit and the downsides of your architecture outweigh it to the point of it being completely negligible.

wow. you folks are either completely clueless or lying through your teeth. neither bodes well for you. there is all the difference in the world between how xen and vmware do things. the large overhead in vmware comes from interpretation of machine code (iow vmware emulates a cpu in software a bit like java does) - xen does not do that.

youre saying this as if it were a bad thing. newsflash: its much, much better. in fact, virtuozzo being a shared kernel solution is why we didnt purchase it. there are numerous resources in the kernel (various datastructures primarily, ie queues) which your virtual environments may have to contend for. you do not and can not provide qos for all of these and even if you wanted to, managing that would be nearly impossible.

the greatest thing about vmware is that vms are almost completely separate - xen gives you that separation, but without vmware's overhead. virtuozzo is just an expensive (albeit technically complex) step up from a grouping fair share process scheduler + chroot jail.

this is a very disingenious way to twist the facts. because virtuozzo runs all environments on the same kernel instance and xen provides separate OS instances, you can physically *boot* less xen environments because each uses more ram.

why is it disingenious then you might ask? because what counts is how many environments you can run under a given load. who the heck is going to run 100 vms? this isnt a mainframe, brother. good luck competing with the z series.

beyond that, what really counts is how much work you can get done per unit of time for your dollar. given that virtuozzo costs only slightly less than a new server, its easy to see how there is absolutely no comparison there.

ive already disproved all of the claims you used to support this assertion. lets look at some benchmarks, xen vs virtuozzo, vs uml, vs vmware. xen beats both uml and vmware hands down. where are the virtuozzo benchmarks?

ahh, thats right - there are none. why? because swsoft makes you sign an nda before you ever see the code, so even if you did benchmarks, youd have to take them to the grave with you. does this sound like a company which is confident in its product? incidentally, vmware's new eula prohibits publishing of comparative benchmarks as well. coincidence? i think not, but its too late now - they got beat ten ways till sunday by a year old alpha version of xen.

i know of enough deployments in mission critical environments to let me sleep well at night. i also know people who run virtuozzo and had spec-fucken-tacular outages with it, not to mention support issues.

you priced yourselves out of the non-enterprise market, im sure your slick fabrications and roi models work well in the enterprise. ive seen them, they dont apply to me and they arent going to apply to anyone on this board, so please spare me.

[quote]
It sounds like you haven't tried Virtuozzo in awhile, it is now available for windows 2003, and if you are interested I can send you the download instructions to test it out.
[quote]

get one of your engineers to answer the disk io question. im interested in technical details. then stop lying. after that, ill consider doing business with you.

look, i realize youre just a sales rep. you sell using the sales material you are given and you dont have the technical background to be able to call bullshit on it. please understand that im not trying to give you personally a hard time, but your product isnt as great as your glossy brochures and koolaid say it is and the dishonesty in your marketing claims would make the guy responsible for the "windows has lower tco than linux" campaign blush.