Maxim Sobolev wrote: > Scott Long wrote: >> F. Senault wrote: >> >>> Monday, May 29, 2006, 10:39:40 AM, you wrote: >>> >>> >>>> Hi, >>> >>> >>>> I wonder if anybody has any objections to importing iSCSI target daemon >>>> from NetBSD (Intel) into the base. >>> >>> >>> Mh, I'm currently doing that, with the help of the author (Alistair G. >>> Crooks), under the form of a port. Alistair just provided me a new >>> version I'm testing, and I was planning to submit the port shortly. >>> (It seems to work quite well). >>> >>> Now, if it's better to include it into the base, so much the better. >>> Alistair was kind enough to take into consideration my suggestions, >>> so, now, the daemon compiles and works under FreeBSD 6 (tested lightly >>> with and i386 and more intensively witn an amd64). >>> >>> The work in progress is here : >>> >>> http://www.lacave.net/~fred/iscsi/ >>> >> >> If it's not going to be integrated into the existing target >> infrastructure then I'd prefer it to be a port. Ultimately it >> would be nice for it to be part of the base system, though. > > Well, arguably we may want to support both ways. Having iSCSI target > running in userland completely has some serious advantages (security is > a big one for example, as you can run daemon easily as unprivileged > process). The kernel iSCSI target only makes sense for really > performance-constrained cases, and hopefully sooner or later we will be > able to narrow the gap by utilizing zero-copy interfaces. P.S. Just to make it clear - just consider running iSCSI over 100MBps link or even a slower WAN links, which I think covers very large market for this technology now. Performance constrain imposed by running in userland is unlikely to be an issue at all. -MaximReceived on Mon May 29 2006 - 20:59:14 UTC
This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:38:56 UTC