[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] Re: [Xen-users] Open source waits for a Xen moment in 2006
On Thu, 2006-01-19 at 21:30 -0500, Costa, Jeff wrote: > Interesting article that talks about this community of Xen > early-adopters: > http://searchopensource.techtarget.com/originalContent/0,289142,sid39_gci1159889,00.html > I just flew over it. I think there's some important and correct information. There's no doubt that the people making xen do a great job and make a good software. Thank you very much! But, to give some hint about what can and should be done better so more people can adopt it, and probably one day also people not into kernel-compiling, c-programming and very low-level system stuff can use xen without killing themselves: As for usability, and the ease of getting started, xen has a long way to go when compared to it's "cousins" like vmware, qemu, and probably the others which I didn't try yet. I just got into it, and I know how to compile a kernel to get the one or other feature, and I have a good basic knowledge of the important network and system setup stuff involved - still it took me much longer to get started with xen than I imagined. Here some details of what I think should be made easier/fixed if the goal is to reach more people, and give them a real easy start with xen: 1) The default kernels are not very well suited for everything I need - they have hardware drivers for everything but the kitchen sink, so the compile takes ages, but they miss basic ip stuff (one example: nfsroot and ip dhcp pnp config from which the nfsroot location could be get cannot be found in the same -xen or -xenU kernel by default together). 2) There are some issues and things which are not well documented yet, or just don't work with some hardware, while it's not defined which hardware has no issues( see my repost an hour ago with acpi issues )and which one has issues. This makes buying hardware for xen an adventure - which private people cannot afford financially, business people just loose their job betting on something which doesn't work with the new hardware they bought for it. 3) The fact that there are multiple different tutorials for the same stuff doesn't make it easier, it just made the choice harder after which tutorial I should go - and most of them don't say for which version of xen they are written. There are also some unclear things in some documentation - in some places it's said one should use the *-xen kernel, in some other places it's said the *-xenU kernel should be used, this information isn't constsistent. Another example of misleading documentation: In the README it's said to compile and install xen like that: # make KERNELS=linux-2.6-xen world # make install Which doesn't work for obviuous reasons, when you got deeper into it, you know it must either be make world make install or make KERNELS=linux-2.6-xen world make KERNELS=linux-2.6-xen install That sounds cheap, but, believe me, for somebody starting with new software it makes adopting xen difficult and feeling of working with flaky, unstable stuff. 4) Apropos unstable - also unclear, is xen 3.0.0 stable or not? If it's stable, why do the distributions packages unpack in xen-unstable? Also sounds small, but for new people it doesn't make a good feeling. Would you feel well in a new car with a sticker on it "this is an unstable crash.test car", even after you made it run 150 miles without crashing it? There'd be always that insecure feeling. If there's some interest I can make a more detailed list of where I saw some errors, I tell them now just by memory of what I saw in the last two weeks. I'd be glad if the xen developers see this message as helpful to improve their software and not as negative criticism. Henning > > ______________________________________________________________________ _______________________________________________ Xen-users mailing list Xen-users@xxxxxxxxxxxxxxxxxxx http://lists.xensource.com/xen-users
|
Lists.xenproject.org is hosted with RackSpace, monitoring our |