<html><body style="word-wrap: break-word; -webkit-nbsp-mode: space; -webkit-line-break: after-white-space; "><div><font class="Apple-style-span" face="Times">The GPFS issue was quite fundamental at the time--IBM would not guarantee that a VM node running GPFS would not corrupt GPFS. They got past the issue eventually, I don't know how.</font></div><div><font class="Apple-style-span" face="Times"><br></font></div><div><font class="Apple-style-span" face="Times">Anyway, we should keep pushing people on this. In the meantime, EC2 and Nimbus are good targets.</font></div><div><br></div><br><div><div>On Apr 24, 2009, at 5:56 PM, Mihael Hategan wrote:</div><br class="Apple-interchange-newline"><blockquote type="cite"><div>On Fri, 2009-04-24 at 17:08 -0500, Ian Foster wrote:<br><blockquote type="cite">It has been a persistent problem getting TG and OSG to consider VMs.<br></blockquote><blockquote type="cite">The reasons are mixed: in my view, some good, some bad. The good<br></blockquote><blockquote type="cite">reasons include the difficulties inherent in getting Xen/VMware<br></blockquote><blockquote type="cite">support for the sometimes odd software found on high-end systems.<br></blockquote><blockquote type="cite">E.g., on TG-UC, both 64-bit hardware and GPFS have been sources of<br></blockquote><blockquote type="cite">problems.<br></blockquote><br>We (swift) can probably plan around GPFS, and possibly other such<br>"difficulties".<br><br><blockquote type="cite"> The bad reason is a persistent conservatism. <br></blockquote><br>Is there any way that the problem can be formulated in such a way as to<br>satisfy both worlds? In my naive view, a node could be used both in a<br>traditional way, and as a VM host when needed.<br><br><br></div></blockquote></div><br></body></html>