Date: Fri, 26 Mar 1999 09:56:31 -0500 (EST) From: "John S. Dyson" <toor@dyson.iquest.net> To: dwilde1@thuntek.net (Donald Wilde) Cc: dyson@iquest.net, advocacy@FreeBSD.ORG Subject: Re: Development Projects (was:Re: FreeBSD emulation for linux) Message-ID: <199903261456.JAA08856@dyson.iquest.net> In-Reply-To: <36FB952B.2E5E7F3C@thuntek.net> from Donald Wilde at "Mar 26, 99 07:09:47 am"
next in thread | previous in thread | raw e-mail | index | archive | help
> > If we can accomplish _just_ these four things with our code-project > resources, plus get a native Oracle 8 port, I think they would give us a > big boost in unit installs. Can I announce the solutions next week? ;-) > I had some internal info about Oracle, and there was probably a predispostion a year or so ago to do a FreeBSD port. However, the previous issues were not technical. (Whatever rants that I have had about lack of FreeBSD marketing quality also apply to me -- so there was little that I could do at Oracle, except to politic those people at NCI who still have contacts there.) To further encourage you, I feel that a bona-fide and competent marketing plan, with realistic estimates might help alot. Also, a comparative estimate of the OS capabilities might show that FreeBSD is quite appropriate for reasonable database support. The version of Oracle running on the NCI stuff was a bit lame, partially because it didn't utilize my AIO stuff (it was done WAY before AIO was written.) From a technical side, FreeBSD is positioned very well to support large scale commercial databases (okay, not large scale, but reasonable department sized applications.) :-). This is one thing that I took very seriously, and part of my longer term plan. Perhaps taking the position that FreeBSD users are often willing to spend money when it is useful for commercial work, and that FreeBSD is a commerce friendly OS. That fact is well known, and large scale and heavily invested organizations are relatively common in the FreeBSD world. The commerce to hacker ratio on FreeBSD is quite large... Hackers don't buy Oracle, but commercial users do. Doing market analysis on a hacker OS, for commercial apps can be very, very tricky. Note my rants about Linux not even supporting large files (and it's historical lack of support for raw files) are from an understanding what DB software really likes to have. Sure the DB software has workarounds for lame OSes, and my AIO code (which worked fine for it's application at Oracle) was part of the attempt to bring FreeBSD up closer to what is really nice for a true medium sized database. (It was also for Oracle video server.) FreeBSD doesn't need threads to support nice RAW AIO behavior. Emulated threading methods just don't work very well from a CPU standpoint, but that emulation was implemented for consistant API reasons. (Doing AIO in libraries is very very lame.) The realtime requirements for many AIO applications are demanding, and libararies that cause more kernel to user transitions are *silly*. If you need technical ammo, I can help you. John To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-advocacy" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199903261456.JAA08856>