![]() | |
#1
| |||
| |||
|
#2
| |||
| |||
|
|
Is there any advantage or disadvantage to going from D3 Linux to D3 Windows? |
#3
| |||
| |||
|
|
"ddspell-m3" wrote: Is there any advantage or disadvantage to going from D3 Linux to D3 Windows? I might be forgetting something but here's my view: snip Does that help? T |
#4
| |||
| |||
|
|
Thanks Tony. What about performance? Which O/S would help D3 achieve better performance? Tony Gravagno wrote: "ddspell-m3" wrote: Is there any advantage or disadvantage to going from D3 Linux to D3 Windows? I might be forgetting something but here's my view: snip Does that help? T |
#5
| |||
| |||
|
|
All things being equal, my instinct says D3Linux would be a faster platform. Then again, I don't know if there are optimizations that I'm unaware of which would allow D3NT to run faster over Win2003. I tend to think of D3NT as being a platform of convenience because it's on the same system as many Windows utilities that I tend to integrate with, and I don't need to have another OS running just to support the DBMS. D3NT can be upgraded on the same OS year after year without changing the OS, whereas D3Linux is built over specific kernels so the OS must be completely reinstalled on an almost yearly basis - that's painful. |
|
It's always better to approach technology from the business perspective. It's like a carpenter saying he needs to secure two beams of a given width - this is a business statement. Compare that to "which is better, nails, staples, or glue?" The same applies for choice of languages, web tools, etc.. (Sorry Chandru) Choose the tool that suits the job that you're going to do. |
#6
| ||||||||||
| ||||||||||
|
|
"Tony Gravagno" wrote All things being equal, my instinct says D3Linux would be a faster platform. Then again, I don't know if there are optimizations that I'm unaware of which would allow D3NT to run faster over Win2003. I tend to think of D3NT as being a platform of convenience because it's on the same system as many Windows utilities that I tend to integrate with, and I don't need to have another OS running just to support the DBMS. D3NT can be upgraded on the same OS year after year without changing the OS, whereas D3Linux is built over specific kernels so the OS must be completely reinstalled on an almost yearly basis - that's painful. |
|
Sorry T, but you're wrong there! I've been running D3 on the same stable kernel and Linux distro for many years now. Actually, I've only upgraded once in the past 8 years. That was from D3/Linux 7.2 to 7.4 for the annoying file flushing bug that was in 7.2. I have no problem running an older kernel, and you know what a tech groupie I am. I prefer to run an older kernel, on enterprise systems, since it has more "miles" on it and I trust that it won't just blindly bomb on me one day. |
|
I can't say that I have been able to use one specific Windows version on other systems for that long of a period. |
|
You only need to upgrade Linux kernels to take advantage of new kernel-only features. |
|
The kernel add-on modules that have newer kernel version requirements are usually things like additional file system capabilities and various new(bleeding edge) hardware technology. On Linux, compiling the system features into the kernel makes the system _much_ faster and also eliminates stability issues with module loading. If you don't want to deal with a kernel upgrade, then don't upgrade the O/S. There is no reason to, unless you have to have capabilities that are only available in a newer kernel. If you want new features in a newer release of D3, which just happens to require a specific kernel for commercial support, then you have no choice but to upgrade the O/S. |
|
That isn't a Linux upgrade or kernel age issue - it's a RD software requirement. |
|
I know many people running D3 on 3 or 4 different distros of Linux with radically different kernel versions. Of course, they don't pay for RD support. D3 is still happy as long as the library dependencies are met and you build a reliable box to run it on. |
|
It's always better to approach technology from the business perspective. It's like a carpenter saying he needs to secure two beams of a given width - this is a business statement. Compare that to "which is better, nails, staples, or glue?" The same applies for choice of languages, web tools, etc.. (Sorry Chandru) Choose the tool that suits the job that you're going to do. I agree with this to some extent. If it is a small installation and a non-techie person is going to be managing the server then I'd suggest a Windows release. If a tech-savvy person is going to be managing it, then the Linux version is going to give you a lot more bang for the buck. You don't need to be a Linux guru to setup and run a Linux server. Linux has come a long way since Slackware 3.0 and its manual configuration steps. |
|
Once properly configured, the server will run by itself for years without any touching. |
|
That's not to say that you shouldn't check on its status regularly and review logs. :P Glen |

#7
| |||
| |||
|
|
Is there any advantage or disadvantage to going from D3 Linux to D3 Windows? Thanks, Danny |
#8
| |||
| |||
|
|
ddspell-m3 wrote: Is there any advantage or disadvantage to going from D3 Linux to D3 Windows? Thanks, Danny What is the motivation for even considering such a move? I'd stay with D3/Linux EVERY time! Whilst D3/Windows 7.5 may address some issues in the future, today D3/Windows lacks features like transaction logging, indices don't work properly in the FSI, the FSI appears to be "fragile", programs can behave differently in the VME & FSI, OSFI has "problems" that don't exist in D3/Linux, the "limited" monitoring capabilities of D3 are even further linited on D3/Windows ... the list goes on. I recently visited a site where I had to reboot their D3/Linux server - uptime was 1473 days ! Tell me again why you are contemplating this move? |
#9
| |||||||||||
| |||||||||||
|
|
Uh, blindly bomb on you like a new and improved Linux upgrade?!? There's an unusual statement. Just kidding, I know what you mean but man is that door wide open... D3 is built and certified on specific kernels, and libs built into distros are built around the kernel. D3 relies on these libs. So as far as D3 is concerned, the underpinnings "should" only be those which RD has built and certified their product to run over. If you have had luck with another distro or a DIY "we don't need no stinkin distro" environment, great for you. But there are no guarantees whatsoever that this will be stable. This is a commercial database, it's not some FC5 hack project. The number of people qualified to build and maintain this environment you speak of are few and far between (kudos to ya bud, seriously!) and running D3 (or most MV DBMS products) on a custom platform is a "run it at your own risk" endeavor. Neither RD nor anyone else (that I know of) will support D3 over a non-certified environment. |
|
I can't say that I have been able to use one specific Windows version on other systems for that long of a period. Your experience and mine balance then. Windows XP came out in 2001 and I haven't used anything else for a desktop since then. That means I've been running D3 7.1.x to 7.4.x on the exact same OS for 5 years. |
|
There are people here who are still running D3NT on the same Windows 2000 that they installed 6 years ago, having upgraded D3 from v7.0 to 7.4.x. I defy anyone to make the same claim for any D3 Linux environment. |

|
You only need to upgrade Linux kernels to take advantage of new kernel-only features. (I'm sorting of going off here but...) Sigh, some applications are compiled using a specific version of GCC and some releases of GCC only work in a given environment. Therefore, if you want to use the programs in question you must be running over the latest kernel. Of course you'll say you can build everything from source, including GCC, for whatever environment you want. There are a tiny handful of people who will even attempt this sort of thing - and for those who have the time, I tip my coffee cup. God help anyone who doesn't know the difference between GCC 2.95.2 and 2.95.3 (yes those are real release numbers) if they're going to do this sort of thing. |
6 to|
Let's not take that 1% and make it seem like this how the normal base of Linux users work. Bottom line for mere mortals is that you're going to upgrade the kernel and everything else when you want some new release of program X. And since most people aren't able to upgrade a Linux box (except for small point releases that come through yum, etc), that means a complete re-install. |
|
The kernel add-on modules that have newer kernel version requirements are usually things like additional file system capabilities and various new(bleeding edge) hardware technology. On Linux, compiling the system features into the kernel makes the system _much_ faster and also eliminates stability issues with module loading. If you don't want to deal with a kernel upgrade, then don't upgrade the O/S. There is no reason to, unless you have to have capabilities that are only available in a newer kernel. If you want new features in a newer release of D3, which just happens to require a specific kernel for commercial support, then you have no choice but to upgrade the O/S. That was my original point. This had nothing to do with Linux on it's own. I was specifically focusing on requirements for D3, not re-hashing the wonders of Linux. For other DBMS products this might not apply, but we aren't talking about others... |
|
That isn't a Linux upgrade or kernel age issue - it's a RD software requirement. Very true. And you've pointed out before that the RedHat distro on which D3 is certified isn't representative of Linux in general. But since D3 is only supported over RH then for this thread I'm sticking to the limitations of this specific environment. I'll maintain that if you try to run a new kernel with old libs or vice versa, all hell breaks loose in RH, and D3 is very sensitive about only running in the certified environment. |
|
I know many people running D3 on 3 or 4 different distros of Linux with radically different kernel versions. Of course, they don't pay for RD support. D3 is still happy as long as the library dependencies are met and you build a reliable box to run it on. Off to the side for a moment, this advice is neither useful nor appropriate for this particular thread. What I mean is we have a noob here who is looking for a platform to run on. If you encourage him to run on something unconventional and unsupported then none of us here will be able to support him and neither will RD. He'll be alone and possibly frustrated with going down the DIY path. In a forum like comp.databases.d3.hack.hack.hack this sort of discussion would be typical but I think we need to stick with what works here without resorting to hardcore hacks. Sometimes what some people Can do shouldn't be confused with advice about what one guy Should be doing. |
|
It's always better to approach technology from the business perspective. It's like a carpenter saying he needs to secure two beams of a given width - this is a business statement. Compare that to "which is better, nails, staples, or glue?" The same applies for choice of languages, web tools, etc.. (Sorry Chandru) Choose the tool that suits the job that you're going to do. I agree with this to some extent. If it is a small installation and a non-techie person is going to be managing the server then I'd suggest a Windows release. If a tech-savvy person is going to be managing it, then the Linux version is going to give you a lot more bang for the buck. You don't need to be a Linux guru to setup and run a Linux server. Linux has come a long way since Slackware 3.0 and its manual configuration steps. That is very true. Once properly configured, the server will run by itself for years without any touching. Until Yum or Apt or Up2Date loads something it shouldn't and your libs are completely screwed and you need to re-install. Sorry dude, it's happened to the last three environments I've installed. Some package erroneously sets up a bad dependency chain and loads something that breaks everything else. Welcome to the wonderful world where a thousand packages are maintained by a thousand different groups of people. I'm tired of it. Can't call me an idiot for what happens here, but these routines are set to run on their own, and they tend to break on their own. My ineptitude is that I can't fix it - though I'd maintain that I shouldn't have to. |
|
That's not to say that you shouldn't check on its status regularly and review logs. :P Glen That's very true too. ![]() Let's not let turn this into a debate about Linux vs Windows. We both have lots of valid arguments and neither side will win. The important thing for now is to just answer Danny's questions about D3 - he can get philosophical rhetoric about operating systems elsewhere. |
|
T |
#10
| |||
| |||
|
|
ddspell-m3 wrote: Is there any advantage or disadvantage to going from D3 Linux to D3 Windows? Thanks, Danny What is the motivation for even considering such a move? I'd stay with D3/Linux EVERY time! Whilst D3/Windows 7.5 may address some issues in the future, today D3/Windows lacks features like transaction logging, indices don't work properly in the FSI, the FSI appears to be "fragile", programs can behave differently in the VME & FSI, OSFI has "problems" that don't exist in D3/Linux, the "limited" monitoring capabilities of D3 are even further linited on D3/Windows ... the list goes on. I recently visited a site where I had to reboot their D3/Linux server - uptime was 1473 days ! Tell me again why you are contemplating this move? |
![]() |
| Thread Tools | |
| Display Modes | |
| |