Re: hardware needed for network stack performance work
(sorry for screwing up the thread; I'm on the daily digest list. Please CC responses to me as well). Hey, I have two machines who are recent, high end, PCI-Express, single CPU dualcore, but unfortunately one's AMD, one's Intel, and I can't find the actual specs as of this moment. I purchased these to experiment with speedy PF firewalls but haven't had the time to actually use them. Also have PCI Express fiber and dual copper syskonnect and intel NICs that go with them. It's a long way to germany from here... If no other company can or will step up, I'll whine, cry, and throw puppydog eyes at my superiors until we can get Henning the hardware he needs, as quickly as possible. I've been occasionally pushing over 1.6 gigabits of transfer over PF recently, and eventually want to start using OpenBGPD routers and the like. I've used many, other hardware platforms for routing and firewalling. They all suck, trust me. We can talk details off list, but I want to throw this out into the open this time. Please, I have no clue how many corporate types would get wind of posts to openbsd-misc, but it's not that big of an investment to send some hardware. It feels real nice to get the right equipment into the right hands. Take your accountant out to a nice lunch, they'll be more understanding of having to do the extra paperwork. In my case, these machines (unfortunately not ideal ones?) are in the wrong hands, my hands, and should probably change. -Dormando
Re: hardware needed for network stack performance work
* Florin Andrei <[EMAIL PROTECTED]> [2007-06-14 02:04]: > Henning Brauer wrote: > > > >Now I am back in Hamburg and would like to continue that work. There > >is quite a lot more performance to gain, but I need to be able to measure, > >profile etc. For that I need two (preferably identical) 1u rackmount, > >very fast single-CPU machines here in Hamburg, asap, since I have some > >time for such development right now. If you can help, please drop deraadt@ > >and me an email. > > I assume the changes that you're making will show up in OpenBSD 4.2? > Or what's the timeframe for including these changes in a "stable" branch > of the code? (i.e. ready for production) if I don't get the proper test hardware the other stuff will rot on my harddisk, probably until c2k8, so they'll then show up in OpenBSD 4.4. I have another ~12.5% (last measurement in calgary) sitting on my disk, almost ready, but can't test properly right now. There's much more to gain. and of course the changes already made will not be in 4.2, they'll be sold extra. sheesh, what do you think we do here? -- Henning Brauer, [EMAIL PROTECTED], [EMAIL PROTECTED] BS Web Services, http://bsws.de Full-Service ISP - Secure Hosting, Mail and DNS Services Dedicated Servers, Rootservers, Application Hosting - Hamburg & Amsterdam
Re: hardware needed for network stack performance work
* Martin Schrvder <[EMAIL PROTECTED]> [2007-06-14 01:24]: > 2007/6/13, Matt Olander <[EMAIL PROTECTED]>: > >Where is Henning located? Shipping free stuff out of country is sometimes a > >pain and takes longer. > > Hamburg, Germany. > > Henning, is DENIC still using OpenBGPD? DECIX, yes, but they're not the right target here, they do their share. -- Henning Brauer, [EMAIL PROTECTED], [EMAIL PROTECTED] BS Web Services, http://bsws.de Full-Service ISP - Secure Hosting, Mail and DNS Services Dedicated Servers, Rootservers, Application Hosting - Hamburg & Amsterdam
Re: hardware needed for network stack performance work
> Henning Brauer wrote: > > > > Now I am back in Hamburg and would like to continue that work. There > > is quite a lot more performance to gain, but I need to be able to measure, > > profile etc. For that I need two (preferably identical) 1u rackmount, > > very fast single-CPU machines here in Hamburg, asap, since I have some > > time for such development right now. If you can help, please drop deraadt@ > > and me an email. > > I assume the changes that you're making will show up in OpenBSD 4.2? > Or what's the timeframe for including these changes in a "stable" branch > of the code? (i.e. ready for production) At the rate that Henning's request is happening, no -- the changes he has not written yet because he has no machines will not make 4.2. In fact, the more discouraged I get with the lack of answers from the community, I start to think that we should slow pf back down again.
Re: hardware needed for network stack performance work
* Florin Andrei <[EMAIL PROTECTED]> [070613 20:02]: > Henning Brauer wrote: > > > >Now I am back in Hamburg and would like to continue that work. There > >is quite a lot more performance to gain, but I need to be able to measure, > >profile etc. For that I need two (preferably identical) 1u rackmount, > >very fast single-CPU machines here in Hamburg, asap, since I have some > >time for such development right now. If you can help, please drop deraadt@ > >and me an email. > > I assume the changes that you're making will show up in OpenBSD 4.2? > Or what's the timeframe for including these changes in a "stable" branch > of the code? (i.e. ready for production) > http://www.openbsd.org/faq/faq1.html#Next
Re: hardware needed for network stack performance work
Henning Brauer wrote: Now I am back in Hamburg and would like to continue that work. There is quite a lot more performance to gain, but I need to be able to measure, profile etc. For that I need two (preferably identical) 1u rackmount, very fast single-CPU machines here in Hamburg, asap, since I have some time for such development right now. If you can help, please drop deraadt@ and me an email. I assume the changes that you're making will show up in OpenBSD 4.2? Or what's the timeframe for including these changes in a "stable" branch of the code? (i.e. ready for production) -- Florin Andrei http://florin.myip.org/
Re: hardware needed for network stack performance work
2007/6/13, Matt Olander <[EMAIL PROTECTED]>: Where is Henning located? Shipping free stuff out of country is sometimes a pain and takes longer. Hamburg, Germany. Henning, is DENIC still using OpenBGPD? Best Martin
Re: hardware needed for network stack performance work
On Wednesday 13 June 2007 10:26 am, Jason Dixon wrote: > On Wed, 13 Jun 2007 11:02:42 -0600, Theo de Raadt <[EMAIL PROTECTED]> wrote: > >> On 6/13/07, Henning Brauer <[EMAIL PROTECTED]> wrote: > >> > very fast single-CPU machines here in Hamburg, asap, since I have some > >> > time for such development right now. If you can help, please drop > > > > deraadt@ > > > >> > and me an email. > >> > >> Got me a t-shirt, a 4.1 CD set, and $100 to you. > > > > Thanks a lot. > > > > However I wish there were some large companies out there using and > > relying in pf, who could just decide (right now) to ship Henning two > > machines. Not because it is the right thing to do, but because they > > will directly benefit, immediately. They could do so completely out > > of self-interest. > > > > But perhaps there are no large companies using pf? That's entirely > > possible, I suppose. > > > > Supporting requests like Henning's out of the pocket change that our > > private user community has is rather crazy; it is a management > > headache for us to wait for money, then move it around. Using such > > (smallish) monies to keep other project needs under control -- that is > > smart. > > > > These performance enhancements will not affect regular private users, > > but will be of particular benefit to companies who use our software in > > larger installs. Companies should stand up when such requests are > > made, or they and their employees should be ashamed of themselves for > > not having any vision, at all. > > I'm probably going to lose a friend over this, but I'd like to challenge > iXsystems to step up and donate a couple systems for this purpose. It > would benefit everyone for you guys to donate the hardware to further > optimize PF. We all know that PF has become as ubiquitous as OpenSSH, at > least in the BSD world. > > How about it Matt, is iXsystems up to the challenge? Damn you. Let me see what I can do. But yes, we had planned to do some OpenBSD support after the BSD Mall merger and integration. What kind of specs are we looking for? And remember, we're not a huge company! I just req'd a few systems for work on FreeBSD 10 GigE support, a system for BSD Cert (plus cash), and I'm sure that I am forgetting a few we recently handed around. Where is Henning located? Shipping free stuff out of country is sometimes a pain and takes longer. We just did a network upgrade, so if you guys need some quality older hardware, we can deliver fairly quickly. P2/P3 Intel based SCSI. There were no problems with the systems, it was just time to upgrade. FYI, we did indeed just switch over to pf on our main gateway. Thank you :-) best, -matt -- Matt Olander CTO, iXsystems - "Servers for Open Source" B http://www.iXsystems.com Public Relations, The FreeBSD Project B B B B http://www.FreeBSD.org BSD on the Desktop! B B B B B B B B B B B B B B http://www.pcbsd.org Phone: (408)943-4100 ext. 113 B B B B B B B B B B Fax: (408)943-4101
Re: hardware needed for network stack performance work
On Wed, 13 Jun 2007 12:01:47 -0600, "Theo de Raadt" <[EMAIL PROTECTED]> said: > > > Oh, a directed spam campaign. perfect. that will endear us to our > > > users. Please return to marketing school from whence you came, and > > > think > > > before you suggest such things. > > > > A open source entity asking for donations from commercial entities > > with whom they already > > have a business relationship is not spam. Theo indicated he wished to > > somehow coalesce and > > redirect appeals so they would result in lump sums and/or machines > > rather than a trickle > > of small user donations. The suggested technique might be effective > > at achieving Theo's > > goals. > > The idea you suggested takes a lot of time and effort. > > We're already spending a lot of time and effort writing things > we give away. > > Noone is going to change their mind based on some campaign. They're > already selfish enough to use free software and believe that there is > no benefit in being the ones to give anything back. Someone else will > give back, and things will keep moving along, that's what they assume. > > No campaign will fix that. > And what are SA's to do? I am so far away from anyone that could cut a check for such a thing it's ridiculous. My managers manager probably couldn't even get such a thing done. I love OBSD, but I also like my job and want to keep it. "BIG" corporations will never be your answer. It will have to come from some medium sized company where the CEO's aren't completely detached from their work force and are technically savvy enough to see how it might benefit them. If anybody knows of anybody like that please chime in. I don't.
Re: hardware needed for network stack performance work
On Wed, 13 Jun 2007, Jason Dixon wrote: I'm probably going to lose a friend over this, but I'd like to challenge iXsystems to step up and donate a couple systems for this purpose. It would benefit everyone for you guys to donate the hardware to further optimize PF. We all know that PF has become as ubiquitous as OpenSSH, at least in the BSD world. How about it Matt, is iXsystems up to the challenge? Thanks, -- Jason Dixon I do know they donated stuff in the past, when I used to buy systems from them. However that's been a number of years. diana
Re: hardware needed for network stack performance work
> Or maybe we need 20 more people like Jason Dixon, to make an appeal to a > company where they have contacts, where the message will at least be > read. That's directly targetted, and therefore more meaningful, and I > think has a higher chance of success. > > Anyone out there know companies using and loving pf? Write them! > Yes this is exactly the intention of the message. we don't want to make the appeal ourselves. Someone on this list has the contacts in places that use it - please make the appeal for us. -Bob
Re: hardware needed for network stack performance work
On 6/13/07, Jack J. Woehr <[EMAIL PROTECTED]> wrote: I agree. I sent two Bluetooth cards and $100 cc donation in the past twelvemonth. right, but it's a $100 per suggestion per email. by my count you're $400 in the hole (-:
Re: hardware needed for network stack performance work
On Wed, Jun 13, 2007 at 12:07:45PM -0600, Jack J. Woehr wrote: > Your point however that it takes work is a justified point. You've > got a www@ responsponsible party. Maybe an appeals@ responsible > party to make webbage, write surveys, and spawn begging campaigns? Or maybe we need 20 more people like Jason Dixon, to make an appeal to a company where they have contacts, where the message will at least be read. That's directly targetted, and therefore more meaningful, and I think has a higher chance of success. Anyone out there know companies using and loving pf? Write them! -- Darrin Chandler| Phoenix BSD User Group | MetaBUG [EMAIL PROTECTED] | http://phxbug.org/ | http://metabug.org/ http://www.stilyagin.com/ | Daemons in the Desert | Global BUG Federation
Re: hardware needed for network stack performance work
> > All fundraising suggestions should be written on the back of a $100 > > bill > > and sent to Theo. > > I agree. I sent two Bluetooth cards and $100 cc donation in the past > twelvemonth. Yes, such small contributions help a lot -- in the places where individuals can help. But when big things are required, it is obvious big players should step up.
Re: hardware needed for network stack performance work
> > No campaign will fix that. > > I dunno, marketing seems to work sometimes. You plain don't get it! You want us to do MORE. We don't want to do more. Keep suggesting it, and I promise we'll do LESS.
Re: hardware needed for network stack performance work
On Jun 13, 2007, at 11:44 AM, Will H. Backman wrote: > All fundraising suggestions should be written on the back of a $100 > bill > and sent to Theo. I agree. I sent two Bluetooth cards and $100 cc donation in the past twelvemonth. -- Jack J. Woehr Director of Development Absolute Performance, Inc. [EMAIL PROTECTED] 303-443-7000 ext. 527
Re: hardware needed for network stack performance work
On Jun 13, 2007, at 12:01 PM, Theo de Raadt wrote: > No campaign will fix that. I dunno, marketing seems to work sometimes. Maybe it can never work *for OpenBSD*, because when some CIO or MIS manager hits the list to ask a question they get roasted by the fachidiot of the day. End of corporate donation. But many non-profit entities do make appeals and campaigns, and it works for them. Your point however that it takes work is a justified point. You've got a www@ responsponsible party. Maybe an appeals@ responsible party to make webbage, write surveys, and spawn begging campaigns? -- Jack J. Woehr Director of Development Absolute Performance, Inc. [EMAIL PROTECTED] 303-443-7000 ext. 527
Re: hardware needed for network stack performance work
> > Oh, a directed spam campaign. perfect. that will endear us to our > > users. Please return to marketing school from whence you came, and > > think > > before you suggest such things. > > A open source entity asking for donations from commercial entities > with whom they already > have a business relationship is not spam. Theo indicated he wished to > somehow coalesce and > redirect appeals so they would result in lump sums and/or machines > rather than a trickle > of small user donations. The suggested technique might be effective > at achieving Theo's > goals. The idea you suggested takes a lot of time and effort. We're already spending a lot of time and effort writing things we give away. Noone is going to change their mind based on some campaign. They're already selfish enough to use free software and believe that there is no benefit in being the ones to give anything back. Someone else will give back, and things will keep moving along, that's what they assume. No campaign will fix that.
Re: hardware needed for network stack performance work
On 6/13/07, Theo de Raadt <[EMAIL PROTECTED]> wrote: These performance enhancements will not affect regular private users, but will be of particular benefit to companies who use our software in larger installs. Companies should stand up when such requests are made, or they and their employees should be ashamed of themselves for not having any vision, at all. That's the problem right there. As a hge user of OpenSSH I'm ashamed of my company for it not stepping up back when I requested donations. It's a lack of vision, pure and simple. Greg -- http://ticketmastersucks.org/tracker.html Dethink to survive - Mclusky
Re: hardware needed for network stack performance work
* Jack J. Woehr <[EMAIL PROTECTED]> [070613 13:27]: > On Jun 13, 2007, at 11:02 AM, Theo de Raadt wrote: > > >However I wish there were some large companies out there using and > >relying in pf, who could just decide (right now) > > Suggestion for tapping the Large Company resource for OpenBSD: > >1) Create an OpenBSD User Survey > a) should include questions that identifies user classes such as > Private Dude and Large Company > b) should allow user to self-identify if willing for > followup surveys and appeals >2) Place survey > a) on website > b) on the next CDROM >3) Use info garnered through survey to > a) craft appeals on website > b) create email appeals to self-identified users in correct > classes. > > Sounds silly perhaps to the more typical OpenBSD user, but if indeed > there is Large Company > use of OpenBSD those admins/users will be more responsive to the > survey-and-appeal paradigm > than our typical lone wolf users. > > -- > Jack J. Woehr > Director of Development > Absolute Performance, Inc. > [EMAIL PROTECTED] > 303-443-7000 ext. 527 > This could possibly be as useful as vendorwatch.org|com whatever it was. And I can say that as I participated in updating it and along with what, three other people.?.?. So in a nutshell, nice idea, but deleted with all the other "Worthless Good Intention Ideas". An admin working in a large company using pf simply needs to pick up the sword and make it happen. Jim
Re: hardware needed for network stack performance work
-Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Theo de Raadt Sent: Wednesday, June 13, 2007 1:30 PM To: Jack J. Woehr Cc: [EMAIL PROTECTED] Org Subject: Re: hardware needed for network stack performance work > On Jun 13, 2007, at 11:02 AM, Theo de Raadt wrote: > > > However I wish there were some large companies out there using and > > relying in pf, who could just decide (right now) > > Suggestion for tapping the Large Company resource for OpenBSD: > > 1) Create an OpenBSD User Survey > a) should include questions that identifies user classes such as > Private Dude and Large Company > b) should allow user to self-identify if willing for > followup surveys and appeals > 2) Place survey > a) on website > b) on the next CDROM > 3) Use info garnered through survey to > a) craft appeals on website > b) create email appeals to self-identified users in correct > classes. > > Sounds silly perhaps to the more typical OpenBSD user, but if indeed > there is Large Company > use of OpenBSD those admins/users will be more responsive to the > survey-and-appeal paradigm > than our typical lone wolf users. All fundraising suggestions should be written on the back of a $100 bill and sent to Theo.
Re: hardware needed for network stack performance work
On Jun 13, 2007, at 11:34 AM, Bob Beck wrote: > > Oh, a directed spam campaign. perfect. that will endear us to our > users. Please return to marketing school from whence you came, and > think > before you suggest such things. A open source entity asking for donations from commercial entities with whom they already have a business relationship is not spam. Theo indicated he wished to somehow coalesce and redirect appeals so they would result in lump sums and/or machines rather than a trickle of small user donations. The suggested technique might be effective at achieving Theo's goals. -- Jack J. Woehr Director of Development Absolute Performance, Inc. [EMAIL PROTECTED] 303-443-7000 ext. 527
Re: hardware needed for network stack performance work
>3) Use info garnered through survey to > a) craft appeals on website Don't need a survey for this. we have a pretty good idea what biggies are using it. > b) create email appeals to self-identified users in correct > classes. Oh, a directed spam campaign. perfect. that will endear us to our users. Please return to marketing school from whence you came, and think before you suggest such things. -Bob
Re: hardware needed for network stack performance work
> On Jun 13, 2007, at 11:02 AM, Theo de Raadt wrote: > > > However I wish there were some large companies out there using and > > relying in pf, who could just decide (right now) > > Suggestion for tapping the Large Company resource for OpenBSD: > > 1) Create an OpenBSD User Survey > a) should include questions that identifies user classes such as > Private Dude and Large Company > b) should allow user to self-identify if willing for > followup surveys and appeals > 2) Place survey > a) on website > b) on the next CDROM > 3) Use info garnered through survey to > a) craft appeals on website > b) create email appeals to self-identified users in correct > classes. > > Sounds silly perhaps to the more typical OpenBSD user, but if indeed > there is Large Company > use of OpenBSD those admins/users will be more responsive to the > survey-and-appeal paradigm > than our typical lone wolf users. That's a great idea. Maybe Henning should write all this stuff up, instead of making pf faster. Let's do more shit-work, so that we have less time to improve the code.
Re: hardware needed for network stack performance work
On Wed, 13 Jun 2007 11:02:42 -0600, Theo de Raadt <[EMAIL PROTECTED]> wrote: >> On 6/13/07, Henning Brauer <[EMAIL PROTECTED]> wrote: >> > very fast single-CPU machines here in Hamburg, asap, since I have some >> > time for such development right now. If you can help, please drop > deraadt@ >> > and me an email. >> >> Got me a t-shirt, a 4.1 CD set, and $100 to you. > > Thanks a lot. > > However I wish there were some large companies out there using and > relying in pf, who could just decide (right now) to ship Henning two > machines. Not because it is the right thing to do, but because they > will directly benefit, immediately. They could do so completely out > of self-interest. > > But perhaps there are no large companies using pf? That's entirely > possible, I suppose. > > Supporting requests like Henning's out of the pocket change that our > private user community has is rather crazy; it is a management > headache for us to wait for money, then move it around. Using such > (smallish) monies to keep other project needs under control -- that is > smart. > > These performance enhancements will not affect regular private users, > but will be of particular benefit to companies who use our software in > larger installs. Companies should stand up when such requests are > made, or they and their employees should be ashamed of themselves for > not having any vision, at all. I'm probably going to lose a friend over this, but I'd like to challenge iXsystems to step up and donate a couple systems for this purpose. It would benefit everyone for you guys to donate the hardware to further optimize PF. We all know that PF has become as ubiquitous as OpenSSH, at least in the BSD world. How about it Matt, is iXsystems up to the challenge? Thanks, -- Jason Dixon DixonGroup Consulting http://www.dixongroup.net
Re: hardware needed for network stack performance work
On Jun 13, 2007, at 11:02 AM, Theo de Raadt wrote: However I wish there were some large companies out there using and relying in pf, who could just decide (right now) Suggestion for tapping the Large Company resource for OpenBSD: 1) Create an OpenBSD User Survey a) should include questions that identifies user classes such as Private Dude and Large Company b) should allow user to self-identify if willing for followup surveys and appeals 2) Place survey a) on website b) on the next CDROM 3) Use info garnered through survey to a) craft appeals on website b) create email appeals to self-identified users in correct classes. Sounds silly perhaps to the more typical OpenBSD user, but if indeed there is Large Company use of OpenBSD those admins/users will be more responsive to the survey-and-appeal paradigm than our typical lone wolf users. -- Jack J. Woehr Director of Development Absolute Performance, Inc. [EMAIL PROTECTED] 303-443-7000 ext. 527
Re: hardware needed for network stack performance work
> On 6/13/07, Henning Brauer <[EMAIL PROTECTED]> wrote: > > very fast single-CPU machines here in Hamburg, asap, since I have some > > time for such development right now. If you can help, please drop deraadt@ > > and me an email. > > Got me a t-shirt, a 4.1 CD set, and $100 to you. Thanks a lot. However I wish there were some large companies out there using and relying in pf, who could just decide (right now) to ship Henning two machines. Not because it is the right thing to do, but because they will directly benefit, immediately. They could do so completely out of self-interest. But perhaps there are no large companies using pf? That's entirely possible, I suppose. Supporting requests like Henning's out of the pocket change that our private user community has is rather crazy; it is a management headache for us to wait for money, then move it around. Using such (smallish) monies to keep other project needs under control -- that is smart. These performance enhancements will not affect regular private users, but will be of particular benefit to companies who use our software in larger installs. Companies should stand up when such requests are made, or they and their employees should be ashamed of themselves for not having any vision, at all.
Re: hardware needed for network stack performance work
On 6/13/07, Henning Brauer <[EMAIL PROTECTED]> wrote: very fast single-CPU machines here in Hamburg, asap, since I have some time for such development right now. If you can help, please drop deraadt@ and me an email. Got me a t-shirt, a 4.1 CD set, and $100 to you. -- "This officer's men seem to follow him merely out of idle curiosity." -- Sandhurst officer cadet evaluation.
hardware needed for network stack performance work
As some of you might have noticed, I worked on network stack and especially pf performance in calgary. This lead to quite massive improvements - one diff in particular doubled pf performance in our test scenario; undeadly covered that: http://www.undeadly.org/cgi?action=article&sid=20070528213858 dlg an I gave a quick talk about it: http://www.openbsd.org/papers/cuug2007/ Now I am back in Hamburg and would like to continue that work. There is quite a lot more performance to gain, but I need to be able to measure, profile etc. For that I need two (preferably identical) 1u rackmount, very fast single-CPU machines here in Hamburg, asap, since I have some time for such development right now. If you can help, please drop deraadt@ and me an email.