Erik, Erik Trimble píše v st 09. 12. 2009 v 06:22 -0800: > Milan Jurik wrote: [...] > > Call those people insane who paid for it. But I met such situations. If > > you have hundreds of gigabytes stored on such system, you are evaluating > > all possible ways. And yes, we did such migration for much smaller price > > than $2k. > > > > How could we? I mean Sun, that is. Base rate for consulting is $100/hr > for short jobs, plus the costs for client acquisition, sales, etc. I > can't see how we didn't lose money. >
Ups, sorry, not we. I, years before I came to Sun. And I know about such situations happening also these days. Yes, no direct money for Sun. [...] > > > > Really? Fine? Are you using it for longer time for data sharing? Not > > just downloading data from USB stick (but even that is not nice). Memory > > hungry, slow etc. > > > That's what Data exchange is: short-term. You don't do high I/O to > data-sharing partitions, as no solution is going to give you reliable, > good performance. I've worked with the ext2 solutions under Windows, > and I would hardly consider them bulletproof or high-speed. Even in > native Linux or *BSD, ext2 is one of the lower-performing filesystems > outside of FAT. > In ext2/ext3 you depend heavilly on way you access the data. Mainly in Linux it depends on tunning/setting. Like in ZFS case. > > > Nobody said something about sharing root partition. But ~/Documents/ on > > dedicated partition? Will you share it on FAT? > > > I tend to mount shared partitions on /shared or something like that, > with my home wherever the local OS decides it should be. /shared is a > FAT partition. You do your work in your normal OS partition, then copy > data over to the shared partition only when you are prepared to reboot > to a new OS. > > Most importantly, these days, dual-boot is a much-less require solution. > Virtualization is a far better choice for a huge variety of places where > dual-boot used to be the only solution. > Not in case of Linux vs. Solaris. Some devices are supported only under Linux. And there are people who prefers running system on bare metal. Because of virtual machines speed. Frequently it is not required, but prefered. > > >> Yes, there are lots of scenarios where you COULD want it. Just like I > >> COULD want to add a jet engine to my Honda. > >> > >> If someone really wants to add ext2/3/4 support, I'm not going to stop > >> them. But I'm certainly not seeing any real use case outside the > >> fringe-hobbyist niche, and nothing that would justify spending any > >> non-volunteer time to support. > >> > > > > Strange that at least once per month we have the same question across > > mailing lists - "where is ext2/ext3 support in Solaris?". Yes, all major > > OSes, including MS Windows with 3rd party solution, has ext2 supported. > > And we use only excuses like "use CIFS/NFS, use FAT". I prefer to spend > > my time to solve it if possible. > > > > Best regards, > > > > Milan > > > > > > The problem is one of target audience. OpenSolaris isn't a wholesale > Linux replacement - that is, it's not intended to do everything that > Linux does, nor be used everywhere that Linux is. People get confused > on that. In many ways, OpenSolaris is more like the various *BSD OSes - > we have a good niche, where our limitations are minimized, and our > strengths reinforced. Staying in niche forced Solaris out of market share. Because niche markets are much smaller these days. Yes, we are able in invade to storage area these days because of ZFS and COMSTAR. But for how long will be ZFS helping us to expand in that market? > People are continually asking for features that > make OSol more like Linux or Windows, which IS THE WRONG THING. > We're NEVER going to have the resources to make OSol a direct competitor > to either in the broad sense. What we do need to spend our time on, > both as a company and as a community is those niches where OpenSolaris > makes sense. When we've fulfilled our feature sets for those niches, > then it's a good idea to look for another niche to expand in. Or, if > someone offers significant financial incentive to expand into a new > niche. But it makes no sense to spread ourselves out, trying to > replicate every feature that someone asks for. > I agree it is important to concentrate on some parts at first. But if somebody is interested in different parts? > For example: over in the ZFS area, you'll see that one of the > longest-asked-for features is the ability of expand a RAID-Z set by > adding a single disk. It's been asked for, repeatedly, by many people, > for over 3 years now. Yet it's down on the priority list of things to > do, for two reasons: (1) it's hard, and thus requires significant > resources to do well, and (2) it's considerably less useful for the ZFS > target audience, which is large-storage systems. > > We need to do the same (and, say the same) to the OpenSolaris community > as a whole: people willing to do the work to scratch their own itch are > more than welcome to add a feature, but the core community needs to > concentrate on our core competencies. > Good. I am not Core contributor to any of our "core competencies". But I am dedicated to support people comming to our system from other systems. Help them with migration. Show them our benefits and lower barriers where it makes sense from my point of view (my point of view because I invest my spare time in it). I went the same path not so long ago. And yes, the project found its users, inside and outside of Sun. Many these things are mainly about students, home offices etc. Places where future decision makers live today. Many times I saw "I will not use Solaris. Yes, ZFS is great, but I will stay with my system because it has - bluetooth, WPA-Enterprise, ext3 support, TV card support etc. Yes, I could use it just for particular thing I do not want to maintain heterogenous enviroment. Linux with ext4 and LVM is just enough for me." So even that does not work. So, ZFS is our excuse sometimes, not benefit. Best regards, Milan _______________________________________________ opensolaris-discuss mailing list [email protected]
