vBulletin Search Engine Optimization
| |||||||
| Register | FAQ | Members List | Calendar | Search | Today's Posts | Mark Forums Read |
| ||||
| Dear gurus, I've recently acquired a couple of near-identical E450s and my colleague is determined to run Linux on one of them. I hear cries of 'bad idea, Solaris will be faster'. And being Solaris biased, I'm inclined to agree. But why, when, for what kind of use? There's quite a bit of conjecture here but very little in the way of fact, at least that I can find. We have an opportunity here to do a bake-off under reasonably controlled conditions. The machines are both E450s, 4x400 Mhz, 2x18 Gb disk, 2048 Mb memory. I'm expecting one to be Solaris 9 or 10 and the other to be based on Gentoo Linux. What kind of tests would the supporters of both camps like us to run, bearing in mind that we don't have pots of money to spend on software such as oracle and will be largely limited to publicly available software or that which comes with the respective OS's? Regards, Pete -- __________________________________________________ __________________ Pete Young pete@antipope.dot.org Remove .dot to reply "Just another crouton, floating on the bouillabaisse of life" |
| |||
| On Mon, 12 Jul 2004 10:05:37 +0100, Pete Young wrote: > Dear gurus, > > I've recently acquired a couple of near-identical E450s and my > colleague is determined to run Linux on one of them. > I hear cries of 'bad idea, Solaris will be faster'. And being > Solaris biased, I'm inclined to agree. > > But why, when, for what kind of use? There's quite a bit of > conjecture here but very little in the way of fact, at least that I > can find. We have an opportunity here to do a bake-off under > reasonably controlled conditions. > > The machines are both E450s, 4x400 Mhz, 2x18 Gb disk, 2048 Mb memory. > I'm expecting one to be Solaris 9 or 10 and the other to be based > on Gentoo Linux. > > What kind of tests would the supporters of both camps like us to > run, bearing in mind that we don't have pots of money to spend on > software such as oracle and will be largely limited to publicly available > software or that which comes with the respective OS's? If you want databases - do use Oracle. Just download it for free from their website, by joining OTN. As long as you use it for testing, devel and such you don't have to pay any licence - you pay only if you use it commercialy. Another good test, imho, would be apache, 2.0 threaded. No point in using a poorly paralelable software on a quad machine. Depending on the enviroment you might want to try some file sharing performance as well... A NFS-ish test, perchance ? Do bear in mind that linux has moved up, and solaris might not prove to be much faster, if any. These machines are low on ram (by todays standards), plus you don't have any storage (which would greatly speed up solaris, while it would remain unsupported on linux...). This is one of the biggest advantages of solaris - support, and reliability. Not speed. Regards, Nikola. |
| |||
| Nikola Krgovic <nkrgovic@sezampro.yu> writes: > Do bear in mind that linux has moved up, and solaris might not prove to >be much faster, if any. These machines are low on ram (by todays standards), >plus you don't have any storage (which would greatly speed up solaris, >while it would remain unsupported on linux...). This is one of the biggest >advantages of solaris - support, and reliability. Not speed. You should not assume that Solaris is stagnant, speed wise. Testing Solaris 10 vs Solaris 9 will show great performance improvements in certain workloads; so testing will show differences there. Casper -- Expressed in this posting are my opinions. They are in no way related to opinions held by my employer, Sun Microsystems. Statements on Sun products included here are not gospel and may be fiction rather than truth. |
| |||
| On Mon, 12 Jul 2004 12:36:32 +0000, Casper H. S. Dik wrote: > Nikola Krgovic <nkrgovic@sezampro.yu> writes: > >> Do bear in mind that linux has moved up, and solaris might not prove to >>be much faster, if any. These machines are low on ram (by todays standards), >>plus you don't have any storage (which would greatly speed up solaris, >>while it would remain unsupported on linux...). This is one of the biggest >>advantages of solaris - support, and reliability. Not speed. > > You should not assume that Solaris is stagnant, speed wise. > > Testing Solaris 10 vs Solaris 9 will show great performance > improvements in certain workloads; so testing will show > differences there. I never did assume so. I just assumed that there are places where a problem was solved differently on linux and solaris. And, that the solaris solution was good and stable, and linux solution was (as it often tends to be) a bit of "cowboy coding", extra fast, non-portable, and a bit not-so-much-stable-but-fast. Some linux developers tend to sacrifice stability and reliability for speed. While this is not a good thing it sometimes shows on tests. Unfortunately it's not possible to test how a system performs after 10 years of uptime Regards, Nikola. |
| |||
| On Mon, 12 Jul 2004 10:05:37 +0100 Pete Young <pete@antipope.dot.org> wrote: > What kind of tests would the supporters of both camps like us to > run, bearing in mind that we don't have pots of money to spend on > software such as oracle and will be largely limited to publicly available > software or that which comes with the respective OS's? Why wouldn't you run tests for software that YOU would run? Personally, I'd love to see an ldap comparison. /fc |
| |||
| LDAP comparison would be lovely, however, I'm afraid such test is limited only to OpenLDAP as I'm not sure iPlanet/SunONE/Java LDAP is available for Linux on SPARC... Frank Cusack wrote: > On Mon, 12 Jul 2004 10:05:37 +0100 Pete Young <pete@antipope.dot.org> wrote: > >>What kind of tests would the supporters of both camps like us to >>run, bearing in mind that we don't have pots of money to spend on >>software such as oracle and will be largely limited to publicly available >>software or that which comes with the respective OS's? > > > Why wouldn't you run tests for software that YOU would run? > > Personally, I'd love to see an ldap comparison. > > /fc |
| ||||
| Frank Cusack <fcusack@fcusack.com> writes: > Why wouldn't you run tests for software that YOU would run? We will, once we decide what we're actually going to do with them. Funny you should mention ldap, that's something we're also interested in. Pete -- __________________________________________________ __________________ Pete Young pete@antipope.dot.org Remove .dot to reply "Just another crouton, floating on the bouillabaisse of life" |