mozilla thunderbird, anyone use it?

[airspirit]

Been a while .... For background, my father in law runs the computer systems for the USN in the NW region + Hawaii, and all of their servers run strictly on *nix platforms. The clients themselves are usually either windows based or are text terminals. As of yet, they haven't had a major breach into their security systems yet, and besides crap like power outages (why they don't have on-site power generation for this while they do for other things is beyond me) and hardware failure they rarely ever have downtime. What I'm hearing from Joe is that because the people already in his department can't handle a change in-house it will cost a ton to switch over ... which is stunningly obvious, because regardless of the platform you switch to if you don't have competent people onboard then you will have to bring in outside help. Don't pretend this is strictly a Linux issue, since you'd have this problem with any other server OS change as well.

As for the email question, I understand that there are more uses for email than strictly text messaging, though many of these "groupware" features are redundancies that can be handled in other ways. Many places are going to secure IM clients to take care of much of this type of work, or other multimedia messaging clients to take care of these needs, which IMHO is an outstanding option for a large and distributed client base.

Hey, I'm not knocking you for using MS ... if it works it works ... what I'm knocking is that you have the blinders on and won't look at the real reasons that you can't switch. Ford Motor Company is going through this same thing right now, having to bring in external experts to customise a Linux distro for use in server and workstation platforms (doing away with the MS products that are causing constant and massive scale shutdowns on mission critical systems ... it is a daily issue). Since all their current (inept ... omg do they suck, which is part of the downtime issue) tech weenies are strictly MS guys, they needed outside support for it. If their tech weenies were already familiar with multiple platforms, however (as they really should be), then that expense (you should see the salary they're offering ... holy sheeit) would be nil.

Just a thought. I'm not a fan of linux for the cost, politics, or anything else: I'm a fan because it works better than the alternative, in my experience.

[Joe]

Originally Posted by airspirit

As for the email question, I understand that there are more uses for email than strictly text messaging, though many of these "groupware" features are redundancies that can be handled in other ways. Many places are going to secure IM clients to take care of much of this type of work, or other multimedia messaging clients to take care of these needs, which IMHO is an outstanding option for a large and distributed client base.

Theres like ummm legal reasons that "IM" and other umm geek technologies arent used in the office. There are "other" solutions... hell going back to tree bark and burnt ends of sticks to write down a calendar is one of them. But all in one package thats works together is far more productive than finding about a dozen apps to do the same job. theres always "other" but doesnt always mean better.

Yes there are no people who specialize in convering a massive 40,000 person network from 100% NT/2K to Linux working for the state. It wasnt just about cost in people, but cost in reaserch, testing, and most packages do cost money to get any kind of support, and really... nothign gets purchased without support tied to it.

The real reasons are quite simple:
1. Linux has no solid or good network security infrastructure without implementing a Novell core for the network.
2. workstation support loads and client support loads would sky rocket trying to do what was multiple domain security into pretty much standalone network security via linux
3. We would loose functionality, predictible upgrade paths, predictible life cycles, and stable product evolution.
4. more specialized staff would be needed (increase programmer/development staff" to support code changes in the many custom systems that would be needed. This is the biggest cost factor of all... going from a system that we pay a low fee to buy and support to now needing people dedicated to supporting that in house.
5. Email systems for Linux are robust in simplistic uses, but lack in almost every other way. Would require a good deal of additional work just to get something that would be able to work with SOME of the specialized applications that use mapi communications to work.
6. Downtime of the conversion would be unavoidable and unacceptable.

Those are the real reasons.

[airspirit]

DAMN YOU I LIKE MY BARK AND STICKS!!!

Er, um ...

There are legal secure IM networks available for inhouse use, though these all have a licensing cost for the server. Nothing wrong there.

As for a predictable upgrade path, that is almost a non-issue in the major projects anymore since they all feed off one another. Upgrading on the right distro is a very simple process, only necessitating a reboot in the case of a kernel version change (which is a rare occurance in the workplace). This is one of the reasons I use Gentoo in-house, since I create the binaries on my box (customized to the meta-distro I've developed off of Gentoo in-house) and then feed from the server down to the workstations seamlessly and transparently. Once done, if there is a change in something like the GUI (only reason for user intervention at all), all you have to do is relog (as opposed to reboot) and you're done. Tech support is also greatly enhanced since I can slip into any machine on my network and make the needed change from my desk rather than hoofing it and inconveniencing the user. This is a particularly beautiful thing, since many of the managers here don't want to be interrupted even if their machine is spouting flames from the floppy drive.

As for security, this can be done very easily if you plan it from the beginning and is more than adequate unless you have a loser setting it up. I really don't understand your reasons there, and since I deal with a smaller and more simplistic network I'm probably not too qualified to argue the point. With that said, I have seen extremely hardened networks that run Linux servers and they are breach-free (like the example I gave above).

What I really wish, however, is perfect emulation of the Windows API and DirectX ... once these things happen, I suspect Windows in all flavors and forms will steadily go the way of the dodo. Unfortunately, I'm not too optimistic it ever will. Since those APIs are closed source, it will be damn near impossible to do a perfect or near-perfect emulation (though look at Samba ... I can always hope). I would love to be able to fire up all of my Windows games in Linux ... I'd never use Windows for anything ever again.

[Joe]

people who want to emulate windows on linux scare me hehehehe

just run windows it does windows emulation better than linux

You upgrade path argument reaks of small network life. Large networks that have set lifecycle/project times, and have many millions of dollars in static budgets for certain projects a given year. Also networks that have essentially 6 different business roles goign on at the same time. You planning 5 years ahead on projects is not that odd these days...

And yes you lack the understanding about the network security control aspect that I am talking about. Account management, centralized client management, pretty much just total overall centralized network/client/authentication control is something linux does not have in any capacity on its own. That alone is why Linux is used in many specific roles but only in very small shops do you see it used as the core infrastructure. really the network as a whole going to Linux would revert it back to a level thats = to that of windows networks back in 1994. Going back in time 10 years just for a free OS

[airspirit]

Mistake on your part: update pushes by servers are done in many places on some massive networks. Since the process can be automated very easily this is a non-issue. It is as simple as having a network mounted package directory (for multiple software platforms you just need to set up a simple database that lists any special needs per machine and/or group) and set up a script that logs to each client in turn and does a push binary install. Of course there are are better ways to do it, but that's a method even a moron couldn't screw up assuming he even wrote the script. One compile (or five if you have an abnormally number of platforms and want to do more than a generic x86 compile) and ten thousand installs ... no problem there. It sure beats the hell out of running windows update on them all and searching for patches for all the individual software products on each individual machine (or using automated update pushes ... which is essentially the same thing, though you're still screwed on the software side).

As for account management and such, have you ever used Linux before? Distributed account management and location transparency is the easiest thing in the world to do. I can log in to my account from any linux box in the building as if I was at my workstation ... no problems there, and I don't have to do anything beyond selecting the username in the menu ... nothing special. I can also manage any other account from anywhere on the network. All you need to do is set up a network mount for your /home directory and you're golden. It would seem you forgot that the *nixes are all natively multiuser and network based operating systems (with the exception of retarded packages like lindows) ....

I don't think you know what you're talking about, or have been advised by ignorant people (microsoft propaganda sheets fall into this category). I'm betting on the latter ... you can't really be held to task if you've been taught [this word's spelling is freaking me out] wrong (which is why I pity the fashionalbe leftists in our colleges rather than drag them down main street by my bumper).

I tell you what, for an obscenely large retainer I'd relocate and hook you guys up, lol.

As for emulating the API: I don't see anything perverse about it. If the software is there, why not make it useful? At least it gives a crossover (much like x86-64 will during the inevitable 64 bit transition) until software houses can start writing for linux. If they wrote for it in the first place there would be no need for emulation at all. In the end, either software houses will magically decide to code for linux while the userbase is low or there will be near-perfect emulation causing the tide to swing to linux that way. Either way would lead to serious problems for Microsoft, though I don't think the former option has a prayer of working (focusing on games in particular).

edit: when I was talking security ignorance, I was referring to the difference in the network protocols and such, not trivial things like account management. I'm sorry I didn't clarify that.

[pHaestus]

I dunno airspirit I kinda equate migrating a larger business environment over to *NIX as being similar to buying AMD-based servers. Sure you can do it, but if something goes awry (and no matter what the platform something will) then it's all because of your decision to go with something unproven (in the eyes of management) and it's your ass. I also think you are trivializing the cost of migration in the form of answering millions of tech support questions every day (how do I burn a cd?) The average business user doesn't want to have to hack text files (xfree86?) any time they change from a larger high res monitor to a smaller LCD and forget to change the res beforehand. I would classify myself as pretty typical of a business power user: I need workstation-class PCs at work and know Win32 pretty well. I am a Linux newbie though with one Mandrake 9.1 box for running code that doesn't work on win32 (and would involve buying programming suites rather than just gcc). I run into TONS of "how do I do this?" questions when using the Linux box on a daily basis, and every one of those questions results in searching the web, installing new packages, and sometimes recompiling. In my life, time is my most valuable asset. The loss of productivity as your whole operation's end users relearn how to use their PCs is IMO never going to make the switch worthwhile. At least for Mandrake 9.1, it's NOT intuitive for Windows users and so what I do now is just remotely connect to the linux box when I need to run any code and don't mess with it.