1. Get DB2, Websphere or Oracle 9i or specifically Oracle Internet Devleoper Suit to install on Fedora... If you can and get Oracle to support it, let me know and I will stop my bitching.
"The upgrade from Fedora to Redhat Enterprise should be quite smooth if you choose to pay for support. But why do you insist that you should have to pay to use up2date, when you can use it for free with Fedora?"
Ok, I want to test switching from Windows to Linux. I need Internet Developer Suite, and some other "linux programs" to work. Ahhh guess what they won't install. I call their support and they say they only support the Enterprise versions. All I want to do is "test" and see "if" I can get off Windows. If I could then I would look at doing it. But guess what? At the current pricing it would cost me significantly more to switch off Windows to Linux.
Ok, a customer comes to us and wants to put up a "help desk" package. It will work on Linux, and they don't have any money. I put up a Fedora box and it works well. They then get some money once it proves itself and the customer now wants to build a "production" box. I then order RedHat E.S. 3.0 and suddenly realize that the kernel is older, and the system is different in many ways AND it cost more than what a Microsoft server would cost.
It use to be that we would go with RedHat because it was far cheaper than Microsoft, we could prototype with it easy and when it went to production we didn't mind paying $350 or so for the "Enterprise" version, but it went from that to $1,500 a YEAR per server. You cannot load that enterprise version on ANY other box. You see my gripe.
"Sounds great, but unfortunately, Redhat doesn't get to decide what other software vendors choose to support, that is up to them. There is nothing preventing these companies from supporting their software on Fedora."
The somewhat do. They could release a version of ES or EWS with no support, that you could load on as many workstations as you want for a small fee. ES could be say $350 and WS could be around $75-100. Notice I said that you could load it as much as you wanted..... At very least allow you to load it on 5-10 machines.
"Tweaked? The GPL is still the GPL as far as I know..."
GPL yes, However RedHat licencing is in addition to GPL.
Lastly, I want to make clear. I am a RedHat fan. I take a LOT of heat from the Microsoft/proprietary fans now that RedHat has increased their pricing by ~10X over the last few years.
RedHat 7.1 (Supported by numerous vendors) - ~$80 - supported for ~3 years. Cost for up2date = $60/year. Total cost for 3 years ~$260
RedHat E.S. ~$800/year. Total cost for 3 year $2,400.
So what has RedHat done to increase the cost by ~10X in 3 years?
"I believe Redhat is also a big contributor to several Free Software projects. They release their improvements under the GPL whenever they can."
Yes, but that doesn't answer my question. How many developers does RedHat employ say vs Microsoft, Oracle, Novell, SAP, etc. Now how many employees does RedHat have?
"I'd demand this from Microsoft as well. I want downloadable ISO version of Windows 2000 Advanced Server. I don't need support. If I ever decide to use Windows Update, then I'll pay."
Microsoft is proprietary and they develop ALL their code. RedHat takes the source for Apache, Samba, VSFTP etc and makes it all work together. In my opinion there is a huge difference.
"What's up with the free stuff? You want free, download Debian and use that. It's pretty stable for enterprise use."
The issue is what RedHat use to do vs what they do now. I don't mind paying, but it should NEVER cost anything close to Windows. Currently in some cases now it cost more. This goes back to my previous point. What does RedHat do? They take other peoples work and package it together. Yes that is a large amount of work, but it is not the same as doing the development.
"And yet you complain about the $5/month fee. I'm confused."
You saw my main complaints. Leasing and enterprise version is different than the workstation. Also, if you read the article, I can't go out and buy 5 versions of this software and pay $25/month. In the article it states that you will have to buy one of their servers and it starts at 10.
"At least you have an alternative. What if you were fed up with Microsoft... you could probably turn to... Microsoft! Oh wow."
We totally agree on this point. If I wasn't a RedHat fan, I probably wouldn't care so much. I just hate to see a company like RedHat abandon the people that made it what it was.
"Market demand decides the price. Corporations see quality support as something worth paying for and will never choose the "we'll get it for free and if something breaks, we'll get a high-school kid to fiddle with the code".
A real world example. I called Redhat with a Apache issue. They had to wait for the "high-school" kids to fix it.
Market demand decides the price. Corporations see quality support as something worth paying for and will never choose the "we'll get it for free and if something breaks, we'll get a high-school kid to fiddle with the code".
Yep, you are correct, and that is why we are now a 50% SuSe shop.
"Business economy 101: Corporations want to always have predictable cash flow. That means that in theory it's much safer to make long-term contracts where the cost is divide over a longer period, because you get rid of the uncertainty over possible major upgrades in the future. Many many companies are moving from buying software to buying services, i.e. "leased software" just because it makes financial sense."
You are correct again, and this will be a factor in going with their competition.
"But if Sun wants to survive, it's time to say bye-bye Solaris."
You and I strongly disagree with this point, I find that there is a significant market for people that want "one throat to choke". So those people love the fact that one vendor does all their stuff. However, you and I do agree that Sun has some serious competion as Linux scales more.
" No business that values its bottom line prices itself like you claim."
I believe Redhat is doing exactly this. Look at where they have come from Redhat 5.x to 7.1. They had a consistant model, and their customers were happy. Now somewhere around 7.2-9.0 they changed and the company has become far more focused on getting money out of their customers as opposed to adding value.
"Maybe RedHat are afraid of looking bad if your supportless platform runs into trouble you can't fix?"
Not sure what you mean by this, but what I want to see from RedHat is a way to purchase the software AND then purchase support on top of that. I want to own the software not lease it, and I want to pay for the support I use. Come up with a price. I also want the ability to load the software on as many machines as I want to.
This is one of the core reasons Linux/RedHat made it in to the enterprise. People like myself would purchase a version of 6.x or 7.x and then load it in an I.T. shop that had a need but didn't want to go through a "formal" I.T. bugeting process. Then people got comfortable with it, then when a new production server was needed, people were more comfortable with RedHat/Linux and they would purchase a version with support. RedHat has just taken that option away from people.
I think what he is saying is, that you could use OpenOffice on Windows, you could use Gimp on Windows etc...
I understand RedHat wants to make money off this deal, and $5 doesn't sound like too much when you factor in that you get support with it, but my two issues are. 1. Redhat takes free software and makes it easy to install, and work with and then charges close to what Microsoft does. I am sorry but how many developers does Redhat have on staff? 2. They want you to basically lease the software from them. That sucks. One good thing with Microsoft windows versions less than XP was that if your company hit hard times you could wait a year or two before an upgrade. Now you will have a fixed cost to Redhat every month.
It appears to me that the bean counters are in charge at RedHat, and they are totally focused on what will make them the most amount of money. Not that this is all bad, but they appear to be doing this at the "cost" of their customers. They kinda seem like another software company I know of.
Redhat, this is what I want. 1. A downloadable ISO version of your enterprise server software, that I can work with but get no support on. I should be able to load this on as many machines as I want to. If I EVER need support on these systems OR want to use up2date on them, then I should have to pay. 2. A desktop version of your software, that is also a free download or a boxed set. Not Fedora!!! This should also have the ability to load on as many computers as I would want to. Again this version would have no support or up2date functionality unless I pay you. This version, unlike Fedora would actually have vendor support from companies like Oracle, Borland, IBM, etc.
The weird part is that I would have little to no problem paying you $100 for your "desktop" system, and I wouldn't mind paying for support afterward. Say $50 a year for up2date services, and support calls at $75 for workstation calls and $300 for server calls.
Either way, Redhat's actions has caused me to start using SuSe, and I suppose in a weird way, I owe you some thanks. I would have NEVER done that before you "tweaked" your licencing.
I agree with your post, and the last thing I want to see is IBM in control of JAVA, well perhaps Microsoft would be worse, but I will defend IBM on one main point. I believe they wrote around 70-80% of the first J2EE standard. So who actually "owns" J2EE? Technically Sun, but IBM spent a lot of time with the code.
IBM's goal is probably not to wrench control of Java away from Sun, but to make a development platform that they fully support and can compete very well with.NET/Microsoft. With that they want to be able to go to a customer and NEVER have to recommend a competitors technology. So no Sun JVM, no Microsoft Client, No Oracle DB, No Apache, No BEA Weblogic, but Linux (IBM/Novell/SuSe), IBM JVM, Websphere, and DB2. Of course this will all be supported on IBM hardware:-) They want hard to be the one stop shop for "ALL" your I.T. needs. They will of course work with any other vendors product...:-)
Lastly, as I told IBM/Novell/Sun/Oracle/BEA/RedHat... your enemy is NOT each other and only together with JAVA can you compete with your true enemy of Microsoft. I think all of these companies except RedHat gets this. For some reason the people I talk to at RedHat seems to hate Sun and anything about them. I had to laugh when I installed my SuSe 9.0 and it had Suns JVM installed by default:-)
Our company tested around 10 JVM's (including different versions) and what we found was the following. 1. Old JVM's were around 10X slower than modern ones. 2. 9 out of 10 JVM's behaved EXACTLY THE SAME!!! Only one JVM had major issues. Any guess which one? Microsofts.... shocker!!!
The Microsoft JVM was fast though, but the latest ones from IBM and Sun were just as fast and behaved correctly.
Also here is the scoop with older to modern JVM's. Sun use to have only a handfull of test to "certify" a JVM. Specifically this was with the 1.0 JVM. Then with 1.1x the test to certify got a little more tough. Now with the 1.2 and above, it got a LOT more tough. I have heard that it went from a few hundred test to many thoughsand test. It is my belief that this coupled with the fact that SWING was decent, made Microsoft aware NOT to support Java at all.
I have developed many JAVA apps, on Linux, Solaris, NetWare, and Windows and have never found any major issues with compatibility with "MY" code. Granted most of my stuff is business apps, so take this with a grain of salt.
I will add that the test that our company did, actually did open up a Microsoft fan boy of their tactics. He tried hard to defend Microsoft, but you can only do so much when EVERY other companies JVM worked.
Now, having seen what Microsoft has been able to do without open-sourcing JAVA, I kind of like the idea of Sun controlling it, in such that they maintain the standard for testing.
Well on our Opteron servers I went with SuSe, because RedHat is way too expensive for 64bit Opteron.
On our Athlons (32bit) we went to Redhat ES 3.0
Here is my limited take on SuSe 64bit and RH 3.0 ES.
SuSe installs easily, their support is great, but you have to submit an email and they usually get back to you within a day. Yast compares well to RedHats up2date. It was a little weird to use ncXXX start instead of service xxx start. What sucks about SuSe 64bit is that a lot of stuff doesn't run well, and for us Oracle 9i is not supported, although I have been told by Oracle that a version of 9i for SuSe 64 bit will be available next month. Now I will add that I rsync between this a SuSe box and an old Redhat 7.1 box, for some reason rsync will eventually hard lock the the SuSe box. This kinda goes with my overall opinion of the stability of SuSe 64bit, things some times lock up.
Now RedHat ES 3.0. Ok this appears to be a lot like Redhat 8.0 and 9.0. If you have worked with those and like them, then you should be happy with ES. Except for the price. On our case it will cost us more to use RedHat than Microsoft Windows 2003 server, and if we were not a Linux shop then we would probably be using Windows. I hope you are hearing that Redhat!!!! You need to offer support contracts and sell ES for around $150. I would have little trouble paying $500 a call for support.
Your post said you are on Redhat 9.0 now. Most major software vendors will NEVER support Redhat 8,9, or Fedora. Given that you were on 9, I will assume that you don't run stuff like Oracle,DB2 or PeopleSoft; so I would evaluate Fedora. I would also take a long look at SuSe 9.0(9.1) non enterprise also. That version will run you under $100 and you could still get support. My only concern is what Novell is going to do to them. Having delt with Novell for over 10 years now, this is something to be very worried about!
Yep and when you go in to do your inital bid for a project, and it cost 10-15X more and requires 2x as many developers AND takes 3X as long... just for an application that will probably only see a lifespan of a year or two: you will realize that "sometimes" it is best to focus on what the customers needs and little more.
I honestly could not see you winning any bids against todays development shops.
Now by no means am I saying always take the easy way out, but I have seen MANY MANY MANY people like you who do "paralysis through analysis" approach, or quite honestly use Rational Rose/RUP approch.
I do agree with some of what you say, but to say that an architect needs to be several years in the future is not accurate in most businesses. Most try and not lock themselves in, but they do not spend needless hours over architecting the most "loosly coupled" or "flexible" system, just for something the customer "might" need in the future. There is a balance between fleibility and time/resources, and I have seen WAY too many so called architects try and build a "super" system, just to never have it compeleted.
I almost think it should be a law that you can't call yourself an architect until you have coded and supported/debuged several systems.
A question I have asked all analyst who interview with me is "How many systems do you have in production now that I can call someone and talk to them about". I have never intervied anyone who has used Rational Rose that has anything in production. I pick on Rose people because they seem to all say the same stuff you said in your previous post... Mostly what I hear is "Company ****** pulled the plug on the project before it was completed".
All that I have learned from application development is that there is no "Silver Bullet". All small to large systems take a ton of work and are a pain to support, and I will agree with you that the more you talk to your customer(s) the better, but talking never builds systems...
I think you miss the point of the previous post. There are many many more developers of open source than any other software company in the world. This includes Microsoft. So why not do both? The people that feel like setting new directions can do so and the people that feel like working on Microsoft stuff can.
Now, personally I think anyone who would consider running C# in a production environment on a Linux box is a bit weird. Generally people that are looking hard at the Microsoft stuff (.net) will run it on Microsoft servers. I haven't talked to one.Net developer that is looking at running any of their stuff on Linux, and more importantly they would not recommend it to their customers. Now the Java people.... that is another story.... they tend to like Linux a lot.
Generally that is a one time charge. Also, I would assume that you mean our finanical department is not in house. If it is then they would be part of the evaluation, and if they believe the company could/would save money, you would be suprised to see how hard they would work:-)
Ok, let me change the question a bit. What is your company going to do when it gets a function that only office 2003 supports and your office 97 can't open it?
Yep, and when you add that new computer to your network, what version of Office is it going to have? This is one of the reasons the guy in the article is looking at OO. Sometimes you are allowed to load an older version of the software, but seeing that Microsoft is now having you register versions online, the day of loading older unsupported software is coming to an end. At least as far as Microsoft is concerned. So yes you are still on office 97, for now.... What will your company do when it starts getting Office 2003 documents and can't open them?
I will add one more thing. I believe that it will only take OpenOffice to get around 25% of the desktop market before compatibility issues start to go away. I believe that at around 25% Microsoft would be forced to make a converter to open OO docs, much like they had to for Word Perfect for years. Just my opinion though... I would just imagine that their customers would start to demand it though...
Unless you go to the company and say. "We currently spend on average $500 a machine for Microsoft Office every three years. We currently have around 300 machines. That is roughly $50,000 a year in hard cost. If we switch off of Microsoft Office to a product that is free and provides comperable features, our company could save that $50,000 a year. In these tough times, we have to seriously consider all options to save cost. With these types of savings it could make it possible to hire an additional person, or in troubled times allow us to keep the people we have. "
Would some people bitch... yep... but most would rather have an additional employee then send that 50k to Redmond EVERY year.
I agree that very little in life is free, but I don't like your reason for possibly staying with Microsoft Office.
Let's take your example. Yes the cost to switch might be more than the upgrade, but you would have to look at the TCO over the long haul. The cost of Microsoft Office in this case will cost that company $$$ every three to five years. This and the fact that they will have no say at all if they want to hold off an upgrade for lets say 6 or 7 years. The company in the article mentioned that this is probably the core reason that they are looking at alternatives to Microsoft Office, they are happy with Office 2k, and don't want to upgrade now, but they have to.
To use another analogy, it is like someone drinking Pepsi in a nice glass, and wanting to switch to drinking water because it is free, but not wanting to because the new glass may cost to much. Yes the one time cost may hurt a little bit, but the long term savings will more than make up for it.
I will agree with you if by saying 99% of the time it is because the administrator didn't have patch xyz installed. Now I want you to realize that there have been over 300Meg of patches released for Windows Server 2000. I believe there are over 40 criticial updates. Come on now... Isn't that a bit extreme?
Now let's say you have a good I.T. shop that wants to have a development, test and production environment. They want all of them to be Windows NT servers. Well it has gotten to the point were that company will need to almost allocate a FTE position just to handle the verification of all these patches and security issues. That is a hard line cost. Or the company can just be like the 100% of the Microsoft shops around here and just throw caution to the wind and just load EVERY patch from Microsoft without testing. Then when something breaks, they call Microsoft and if it isn't a Microsoft product, they seem to always get "Well you should be running our product for that".
That's great for you, but my Apache logs show a different story for Windows users. We also have not had any major outbreaks in a while. However, we have had MAJOR issues with some of the hot fixes and service packs that have been released. Also, have you noticed the size of these things lately? 100 Megs!!!
Now let me say that our Macintosh and Linux clients have NEVER had a virus issue at all. To be honest we don't have to spend time with virus software for Linux at all. So no more testing all the definition updates to see if they hose out your system. I assume, being a good windows administrator you do test all of the latest virus definitions before you let them loose on your network... given that they come out about every 4 days or so, that time in testing sure needs to be counted in the TCO.
Not to mention the cost associated with Norton or other supported virus software.
The TCO of Windows is not small. I am not saying Linux is better for you, but for us we are looking very seriously at migrating. To be honest if Apple released OSX for x86 we would probably move in that direction, but given that isn't going to happen... Linux is the next best choice for us.
Currently we have only one small issue that has been found and one package that we are looking at a replacement for. The new Oracle reports will not run. When we tried this evaluation a year and a half ago, there were a TON of things that was not workable. It is my belief that "if" we can get Oracle forms working, then this project will move to the next phase. Does that mean it will happen? Nope, but if it doesn't, I will bet you that given the rapid development of good software for Linux, our company will be taking another look at it in a year or two.
Ahh but the reverse is true. If you are the administrator of a small company, like I am and you run Windows XP on your clients, like I do and they get hit with some new worm that cost your business a couple of days of downtime, you have to be prepared to take the heat on that one.
We are in the process of evaluating switching some of our clients to Linux. I am not going to say it WILL happen, but it is being evaluated. I would assume that any company that is looking at switching would also go through a similar process. If they do a good job then most of the issues will be flushed out.
Are you talking about a Jtable? If so, I understand that it isn't easy to work with a Jtable or a Jtree. Both are somewhat complex. However, like all things that complexity comes with some power. If I remember the issue is that it uses a very strick model view controller pattern.
I am more curious what objects you had to extend in SWING that you don't in.Net.
Can you provide some examples? One I can think that use to exist was the inability to limit text input on a jtextfield. However, as you mentioned, you extend it one time and then you are done.
Good point. Sorry about that. I have been a bit jaded about the whole Microsoft thing because of their tatics inside companies that I have worked for. That is another story though...
I apologize for any attack on you.
I still would like to know what inside of.Net makes codeing better than JAVA? Let's not debate the GUI. I think we will agree that good GUI development is hard either way.
You don't like SWING... you say it is because SUN made it cross platform. Weird, I worked with SWING a ton for two years, and found it ok. There were things I didn't like about it, but in general I liked it. The learning curve was a bit steep, but once done it wan't bad. I personally love the fact that if I write an application on Linux (and don't use an X,Y, layout manager) it will look and feel good on a Mac, NetWare, Solaris, and even my PDA. No it doesn't use native widgets, but because of that it doesn't behave weird on different platforms. Do I have some issues with SWING? Yep, but all in all it works fine.
When is the last time I saw a XYZ app done as a web application? Off the top of my head I can think about turbo tax. But in general I agree with you. However, I also believe that will change over time. The "idea" of applets will come again. People are sick and tired of paying for software they take home, use once in a blue moon and then have it crash their system after an upgrade or worse yet require some other piece of software to be loaded first.
You say.Net is a better desktop app environment. How? What specifically is better? I am not trying to be a jerk here, just to understand it. I would think tools such as webstart would be something you would like.
A Java CPU... Why? The only issue I have with Java is memory usage, not CPU. On a ton of test JAVA has run neck and neck with C. That isn't too bad...
Open source Java - Yep that would be bad, Microsoft would grab it and change it/criple it and then put it on EVERY version of Windows from now on. Then people like yourself would say "We have to code to Microsoft's JVM... it is the defacto standard". I have been there and done that. I am so glad Microsoft is out of the Java business.
I agree with a lot you said, and again I am not trying to be a jerk here.
It is funny you mentioned SWING. I spent two years of my life developing SWING apps, and found the learning curve to be quite large, but it works great. I also didn't come from an object oriented background and that made learning tough. So would I say that SWING is better on Windows than Windows GDI? Nope, but would I say it is good enough for a very large percentage of applications? Yes. An issue I see with Windows development as the GUI goes is what version/service pack are you coding for. A lot of people still have a mix of 95,98,98SE,NT 4.0,2000 and XP. Yes there are similarities, but there are a ton of issues. This is were ideas like Java's WebStart start to shine. You as a developer have some control on what the client is running. As I mentioned some smaller shops can control what versions of clients they have, but it gets almost impossible at the larger shops (Dow Chemical being one execption, but their standards have caused them to delay upgrades for years). So yes their developers have a standard, but that standard is old. If it isn't old, then it will be in a year...
Now, good gui development isn't easy, well I should say it isn't easy for me:-) Getting windows to scale and run well (threaded), takes some good design. I have found SWING to be able to handle most things well, granted there were some issue, but there were issues with VB.
So yes 99% of the people do have a Windows box at work. I agree. Will you agree that of that 99% no more than 25% are running the same version of Windows? I won't even count the numerous service packs that make apps behave differently. (Like most even numbered service packs:-)
Now for SQL server V.S. Oracle. That was what the company wanted to use. Oracle isn't cheap, but neither is SQL server. Also, I just hired a good Oracle DBA. They are the SAME cost as a good SQL server DBA. (We use to have SQL server) (I live in Indiana, labor is very very cheap at the moment, we are a manufacturing state and have been hit hard by the recent economy). I also have worked with DB2 and I agree with what you said about not discounting any DB. This includes of course PostgreSQL and MySQL.
Lastly, I think you and I agree on a lot, I just want to make a few points. 1. Microsoft will NEVER NEVER allow.Net to run well on anything but Windows. 2. If/when Microsoft looses marketshare that code will become a liability if it needs to be ported. 3. Competition is great. It makes everyone better.
Ok, since this is getting long, I add one more point; and pick on Oracle. Oracle has a development package called Jdeveloper. It can use an object to relational mapping tool called BC4J. Once you have your BC4J objects you can call them via their front end tools (UIX controls). This looks great on the surface. I was able to bang out an applicaton in 50% of the time it took previously. Like you I do a traditional development style. This tool looked to be the saving grace, and our estimates for projects looked like they could drop by a very large margin. Ahh but then the issue came in. I won't go in to the HUGE list of issue, but suffice to say we will not be using their stuff for development for a long while. This is exactly how I feel about.NET. It isn't bad "if" you know what you are getting in for, but it will probably come back and haunt you in the future.
If you worked with JSP then you know of things like custom tags. In my opinion JSP is quite a bit ahead of ASP.
You like Javascript the best? Wow you are the only person I have talked/written to that says that. Weird. Most people hate it.
If you believe that Mono will ever bring quality C sharp apps to Linux then you better start learning another language. Name an instance in Microsofts past that shows how they have worked with another vendor to run their apps on a non Microsoft platform. I am glad that these people want to develop Mono, but to be honest I can never see any I.T. shop that is doing.net stuff use anything but Microsoft. The same isn't said for C or Java.
Again, if you like.net then you need to seriously look at Java again. It does everthing, and then some(container managed entity beans), that.net does. The core difference is that it doesn't lock you in to Microsoft.
I am not trying to be a jerk here. I just want to warn you that having worked with Microsoft for a while, they will make things easy at the cost of security and scale. They will NEVER allow their cash cow Windows to be jeprodised. Their apps tend not to scale well. If you want I can point you to some shops that are probably going to close soon that bet the farm with Microsoft. One in particular has had the boys from Redmond in for a while and this was some of the recommendations. 1. Reboot your servers every night. Granted this company has over 400 Win2k boxes. 2. Buy more hardware. Yet people like me have shown that you could scale to what they want with some effort using non Microsoft.
The last one kills me.
3. Get rid of Oracle and go with SQL server. Yet their DB and connection is not a factor at all. This is the ONLY non Microsoft product they own, and yet Microsoft "techs" want it ripped out.
I can name many other companies if you like. I don't want to say Microsoft won't work at all, it will, and in some smaller shops it might work well. I just want to warn people what they are getting in to before they spend their hard earned money.
Please Please Please take from this that.Net will never work well on any platform other than Windows. Microsoft will make sure of that.
Slashdot Mirror
1. Get DB2, Websphere or Oracle 9i or specifically Oracle Internet Devleoper Suit to install on Fedora... If you can and get Oracle to support it, let me know and I will stop my bitching.
"The upgrade from Fedora to Redhat Enterprise should be quite smooth if you choose to pay for support. But why do you insist that you should have to pay to use up2date, when you can use it for free with Fedora?"
Ok, I want to test switching from Windows to Linux. I need Internet Developer Suite, and some other "linux programs" to work. Ahhh guess what they won't install. I call their support and they say they only support the Enterprise versions. All I want to do is "test" and see "if" I can get off Windows. If I could then I would look at doing it. But guess what? At the current pricing it would cost me significantly more to switch off Windows to Linux.
Ok, a customer comes to us and wants to put up a "help desk" package. It will work on Linux, and they don't have any money. I put up a Fedora box and it works well. They then get some money once it proves itself and the customer now wants to build a "production" box. I then order RedHat E.S. 3.0 and suddenly realize that the kernel is older, and the system is different in many ways AND it cost more than what a Microsoft server would cost.
It use to be that we would go with RedHat because it was far cheaper than Microsoft, we could prototype with it easy and when it went to production we didn't mind paying $350 or so for the "Enterprise" version, but it went from that to $1,500 a YEAR per server. You cannot load that enterprise version on ANY other box. You see my gripe.
"Sounds great, but unfortunately, Redhat doesn't get to decide what other software vendors choose to support, that is up to them. There is nothing preventing these companies from supporting their software on Fedora."
The somewhat do. They could release a version of ES or EWS with no support, that you could load on as many workstations as you want for a small fee. ES could be say $350 and WS could be around $75-100. Notice I said that you could load it as much as you wanted..... At very least allow you to load it on 5-10 machines.
"Tweaked? The GPL is still the GPL as far as I know..."
GPL yes, However RedHat licencing is in addition to GPL.
Lastly, I want to make clear. I am a RedHat fan. I take a LOT of heat from the Microsoft/proprietary fans now that RedHat has increased their pricing by ~10X over the last few years.
RedHat 7.1 (Supported by numerous vendors) - ~$80 - supported for ~3 years. Cost for up2date = $60/year. Total cost for 3 years ~$260
RedHat E.S. ~$800/year. Total cost for 3 year $2,400.
So what has RedHat done to increase the cost by ~10X in 3 years?
"I believe Redhat is also a big contributor to several Free Software projects. They release their improvements under the GPL whenever they can."
Yes, but that doesn't answer my question. How many developers does RedHat employ say vs Microsoft, Oracle, Novell, SAP, etc. Now how many employees does RedHat have?
"I'd demand this from Microsoft as well. I want downloadable ISO version of Windows 2000 Advanced Server. I don't need support. If I ever decide to use Windows Update, then I'll pay."
Microsoft is proprietary and they develop ALL their code. RedHat takes the source for Apache, Samba, VSFTP etc and makes it all work together. In my opinion there is a huge difference.
"What's up with the free stuff? You want free, download Debian and use that. It's pretty stable for enterprise use."
The issue is what RedHat use to do vs what they do now. I don't mind paying, but it should NEVER cost anything close to Windows. Currently in some cases now it cost more. This goes back to my previous point. What does RedHat do? They take other peoples work and package it together. Yes that is a large amount of work, but it is not the same as doing the development.
"And yet you complain about the $5/month fee. I'm confused."
You saw my main complaints. Leasing and enterprise version is different than the workstation. Also, if you read the article, I can't go out and buy 5 versions of this software and pay $25/month. In the article it states that you will have to buy one of their servers and it starts at 10.
"At least you have an alternative. What if you were fed up with Microsoft... you could probably turn to... Microsoft! Oh wow."
We totally agree on this point. If I wasn't a RedHat fan, I probably wouldn't care so much. I just hate to see a company like RedHat abandon the people that made it what it was.
"Market demand decides the price. Corporations see quality support as something worth paying for and will never choose the "we'll get it for free and if something breaks, we'll get a high-school kid to fiddle with the code".
A real world example. I called Redhat with a Apache issue. They had to wait for the "high-school" kids to fix it.
Market demand decides the price. Corporations see quality support as something worth paying for and will never choose the "we'll get it for free and if something breaks, we'll get a high-school kid to fiddle with the code".
Yep, you are correct, and that is why we are now a 50% SuSe shop.
"Business economy 101: Corporations want to always have predictable cash flow. That means that in theory it's much safer to make long-term contracts where the cost is divide over a longer period, because you get rid of the uncertainty over possible major upgrades in the future. Many many companies are moving from buying software to buying services, i.e. "leased software" just because it makes financial sense."
You are correct again, and this will be a factor in going with their competition.
"But if Sun wants to survive, it's time to say bye-bye Solaris."
You and I strongly disagree with this point, I find that there is a significant market for people that want "one throat to choke". So those people love the fact that one vendor does all their stuff. However, you and I do agree that Sun has some serious competion as Linux scales more.
" No business that values its bottom line prices itself like you claim."
I believe Redhat is doing exactly this. Look at where they have come from Redhat 5.x to 7.1. They had a consistant model, and their customers were happy. Now somewhere around 7.2-9.0 they changed and the company has become far more focused on getting money out of their customers as opposed to adding value.
"Maybe RedHat are afraid of looking bad if your supportless platform runs into trouble you can't fix?"
Not sure what you mean by this, but what I want to see from RedHat is a way to purchase the software AND then purchase support on top of that. I want to own the software not lease it, and I want to pay for the support I use. Come up with a price. I also want the ability to load the software on as many machines as I want to.
This is one of the core reasons Linux/RedHat made it in to the enterprise. People like myself would purchase a version of 6.x or 7.x and then load it in an I.T. shop that had a need but didn't want to go through a "formal" I.T. bugeting process. Then people got comfortable with it, then when a new production server was needed, people were more comfortable with RedHat/Linux and they would purchase a version with support. RedHat has just taken that option away from people.
I think what he is saying is, that you could use OpenOffice on Windows, you could use Gimp on Windows etc...
I understand RedHat wants to make money off this deal, and $5 doesn't sound like too much when you factor in that you get support with it, but my two issues are.
1. Redhat takes free software and makes it easy to install, and work with and then charges close to what Microsoft does. I am sorry but how many developers does Redhat have on staff?
2. They want you to basically lease the software from them. That sucks. One good thing with Microsoft windows versions less than XP was that if your company hit hard times you could wait a year or two before an upgrade. Now you will have a fixed cost to Redhat every month.
It appears to me that the bean counters are in charge at RedHat, and they are totally focused on what will make them the most amount of money. Not that this is all bad, but they appear to be doing this at the "cost" of their customers. They kinda seem like another software company I know of.
Redhat, this is what I want.
1. A downloadable ISO version of your enterprise server software, that I can work with but get no support on. I should be able to load this on as many machines as I want to. If I EVER need support on these systems OR want to use up2date on them, then I should have to pay.
2. A desktop version of your software, that is also a free download or a boxed set. Not Fedora!!! This should also have the ability to load on as many computers as I would want to. Again this version would have no support or up2date functionality unless I pay you. This version, unlike Fedora would actually have vendor support from companies like Oracle, Borland, IBM, etc.
The weird part is that I would have little to no problem paying you $100 for your "desktop" system, and I wouldn't mind paying for support afterward. Say $50 a year for up2date services, and support calls at $75 for workstation calls and $300 for server calls.
Either way, Redhat's actions has caused me to start using SuSe, and I suppose in a weird way, I owe you some thanks. I would have NEVER done that before you "tweaked" your licencing.
I agree with your post, and the last thing I want to see is IBM in control of JAVA, well perhaps Microsoft would be worse, but I will defend IBM on one main point. I believe they wrote around 70-80% of the first J2EE standard. So who actually "owns" J2EE? Technically Sun, but IBM spent a lot of time with the code.
.NET/Microsoft. With that they want to be able to go to a customer and NEVER have to recommend a competitors technology. So no Sun JVM, no Microsoft Client, No Oracle DB, No Apache, No BEA Weblogic, but Linux (IBM/Novell/SuSe), IBM JVM, Websphere, and DB2. Of course this will all be supported on IBM hardware :-) They want hard to be the one stop shop for "ALL" your I.T. needs. They will of course work with any other vendors product... :-)
:-)
IBM's goal is probably not to wrench control of Java away from Sun, but to make a development platform that they fully support and can compete very well with
Lastly, as I told IBM/Novell/Sun/Oracle/BEA/RedHat... your enemy is NOT each other and only together with JAVA can you compete with your true enemy of Microsoft. I think all of these companies except RedHat gets this. For some reason the people I talk to at RedHat seems to hate Sun and anything about them. I had to laugh when I installed my SuSe 9.0 and it had Suns JVM installed by default
Here is my 2c....
Our company tested around 10 JVM's (including different versions) and what we found was the following.
1. Old JVM's were around 10X slower than modern ones.
2. 9 out of 10 JVM's behaved EXACTLY THE SAME!!! Only one JVM had major issues. Any guess which one? Microsofts.... shocker!!!
The Microsoft JVM was fast though, but the latest ones from IBM and Sun were just as fast and behaved correctly.
Also here is the scoop with older to modern JVM's. Sun use to have only a handfull of test to "certify" a JVM. Specifically this was with the 1.0 JVM. Then with 1.1x the test to certify got a little more tough. Now with the 1.2 and above, it got a LOT more tough. I have heard that it went from a few hundred test to many thoughsand test. It is my belief that this coupled with the fact that SWING was decent, made Microsoft aware NOT to support Java at all.
I have developed many JAVA apps, on Linux, Solaris, NetWare, and Windows and have never found any major issues with compatibility with "MY" code. Granted most of my stuff is business apps, so take this with a grain of salt.
I will add that the test that our company did, actually did open up a Microsoft fan boy of their tactics. He tried hard to defend Microsoft, but you can only do so much when EVERY other companies JVM worked.
Now, having seen what Microsoft has been able to do without open-sourcing JAVA, I kind of like the idea of Sun controlling it, in such that they maintain the standard for testing.
Well on our Opteron servers I went with SuSe, because RedHat is way too expensive for 64bit Opteron.
On our Athlons (32bit) we went to Redhat ES 3.0
Here is my limited take on SuSe 64bit and RH 3.0 ES.
SuSe installs easily, their support is great, but you have to submit an email and they usually get back to you within a day. Yast compares well to RedHats up2date. It was a little weird to use ncXXX start instead of service xxx start. What sucks about SuSe 64bit is that a lot of stuff doesn't run well, and for us Oracle 9i is not supported, although I have been told by Oracle that a version of 9i for SuSe 64 bit will be available next month. Now I will add that I rsync between this a SuSe box and an old Redhat 7.1 box, for some reason rsync will eventually hard lock the the SuSe box. This kinda goes with my overall opinion of the stability of SuSe 64bit, things some times lock up.
Now RedHat ES 3.0. Ok this appears to be a lot like Redhat 8.0 and 9.0. If you have worked with those and like them, then you should be happy with ES. Except for the price. On our case it will cost us more to use RedHat than Microsoft Windows 2003 server, and if we were not a Linux shop then we would probably be using Windows. I hope you are hearing that Redhat!!!! You need to offer support contracts and sell ES for around $150. I would have little trouble paying $500 a call for support.
Your post said you are on Redhat 9.0 now. Most major software vendors will NEVER support Redhat 8,9, or Fedora. Given that you were on 9, I will assume that you don't run stuff like Oracle,DB2 or PeopleSoft; so I would evaluate Fedora. I would also take a long look at SuSe 9.0(9.1) non enterprise also. That version will run you under $100 and you could still get support. My only concern is what Novell is going to do to them. Having delt with Novell for over 10 years now, this is something to be very worried about!
Kinda like they did with SARS.
Yep and when you go in to do your inital bid for a project, and it cost 10-15X more and requires 2x as many developers AND takes 3X as long... just for an application that will probably only see a lifespan of a year or two: you will realize that "sometimes" it is best to focus on what the customers needs and little more.
I honestly could not see you winning any bids against todays development shops.
Now by no means am I saying always take the easy way out, but I have seen MANY MANY MANY people like you who do "paralysis through analysis" approach, or quite honestly use Rational Rose/RUP approch.
I do agree with some of what you say, but to say that an architect needs to be several years in the future is not accurate in most businesses. Most try and not lock themselves in, but they do not spend needless hours over architecting the most "loosly coupled" or "flexible" system, just for something the customer "might" need in the future. There is a balance between fleibility and time/resources, and I have seen WAY too many so called architects try and build a "super" system, just to never have it compeleted.
I almost think it should be a law that you can't call yourself an architect until you have coded and supported/debuged several systems.
A question I have asked all analyst who interview with me is "How many systems do you have in production now that I can call someone and talk to them about". I have never intervied anyone who has used Rational Rose that has anything in production. I pick on Rose people because they seem to all say the same stuff you said in your previous post... Mostly what I hear is "Company ****** pulled the plug on the project before it was completed".
All that I have learned from application development is that there is no "Silver Bullet". All small to large systems take a ton of work and are a pain to support, and I will agree with you that the more you talk to your customer(s) the better, but talking never builds systems...
I think you miss the point of the previous post. There are many many more developers of open source than any other software company in the world. This includes Microsoft. So why not do both? The people that feel like setting new directions can do so and the people that feel like working on Microsoft stuff can.
.Net developer that is looking at running any of their stuff on Linux, and more importantly they would not recommend it to their customers. Now the Java people.... that is another story.... they tend to like Linux a lot.
Now, personally I think anyone who would consider running C# in a production environment on a Linux box is a bit weird. Generally people that are looking hard at the Microsoft stuff (.net) will run it on Microsoft servers. I haven't talked to one
Generally that is a one time charge. Also, I would assume that you mean our finanical department is not in house. If it is then they would be part of the evaluation, and if they believe the company could/would save money, you would be suprised to see how hard they would work :-)
Ok, let me change the question a bit. What is your company going to do when it gets a function that only office 2003 supports and your office 97 can't open it?
Yep, and when you add that new computer to your network, what version of Office is it going to have? This is one of the reasons the guy in the article is looking at OO. Sometimes you are allowed to load an older version of the software, but seeing that Microsoft is now having you register versions online, the day of loading older unsupported software is coming to an end. At least as far as Microsoft is concerned. So yes you are still on office 97, for now.... What will your company do when it starts getting Office 2003 documents and can't open them?
I will add one more thing. I believe that it will only take OpenOffice to get around 25% of the desktop market before compatibility issues start to go away. I believe that at around 25% Microsoft would be forced to make a converter to open OO docs, much like they had to for Word Perfect for years. Just my opinion though... I would just imagine that their customers would start to demand it though...
Unless you go to the company and say.
"We currently spend on average $500 a machine for Microsoft Office every three years. We currently have around 300 machines. That is roughly $50,000 a year in hard cost. If we switch off of Microsoft Office to a product that is free and provides comperable features, our company could save that $50,000 a year. In these tough times, we have to seriously consider all options to save cost. With these types of savings it could make it possible to hire an additional person, or in troubled times allow us to keep the people we have. "
Would some people bitch... yep... but most would rather have an additional employee then send that 50k to Redmond EVERY year.
I agree that very little in life is free, but I don't like your reason for possibly staying with Microsoft Office.
Let's take your example. Yes the cost to switch might be more than the upgrade, but you would have to look at the TCO over the long haul. The cost of Microsoft Office in this case will cost that company $$$ every three to five years. This and the fact that they will have no say at all if they want to hold off an upgrade for lets say 6 or 7 years. The company in the article mentioned that this is probably the core reason that they are looking at alternatives to Microsoft Office, they are happy with Office 2k, and don't want to upgrade now, but they have to.
To use another analogy, it is like someone drinking Pepsi in a nice glass, and wanting to switch to drinking water because it is free, but not wanting to because the new glass may cost to much. Yes the one time cost may hurt a little bit, but the long term savings will more than make up for it.
I will agree with you if by saying 99% of the time it is because the administrator didn't have patch xyz installed. Now I want you to realize that there have been over 300Meg of patches released for Windows Server 2000. I believe there are over 40 criticial updates. Come on now... Isn't that a bit extreme?
Now let's say you have a good I.T. shop that wants to have a development, test and production environment. They want all of them to be Windows NT servers. Well it has gotten to the point were that company will need to almost allocate a FTE position just to handle the verification of all these patches and security issues. That is a hard line cost. Or the company can just be like the 100% of the Microsoft shops around here and just throw caution to the wind and just load EVERY patch from Microsoft without testing. Then when something breaks, they call Microsoft and if it isn't a Microsoft product, they seem to always get "Well you should be running our product for that".
That's great for you, but my Apache logs show a different story for Windows users. We also have not had any major outbreaks in a while. However, we have had MAJOR issues with some of the hot fixes and service packs that have been released. Also, have you noticed the size of these things lately? 100 Megs!!!
Now let me say that our Macintosh and Linux clients have NEVER had a virus issue at all. To be honest we don't have to spend time with virus software for Linux at all. So no more testing all the definition updates to see if they hose out your system. I assume, being a good windows administrator you do test all of the latest virus definitions before you let them loose on your network... given that they come out about every 4 days or so, that time in testing sure needs to be counted in the TCO.
Not to mention the cost associated with Norton or other supported virus software.
The TCO of Windows is not small. I am not saying Linux is better for you, but for us we are looking very seriously at migrating. To be honest if Apple released OSX for x86 we would probably move in that direction, but given that isn't going to happen... Linux is the next best choice for us.
Currently we have only one small issue that has been found and one package that we are looking at a replacement for. The new Oracle reports will not run. When we tried this evaluation a year and a half ago, there were a TON of things that was not workable. It is my belief that "if" we can get Oracle forms working, then this project will move to the next phase. Does that mean it will happen? Nope, but if it doesn't, I will bet you that given the rapid development of good software for Linux, our company will be taking another look at it in a year or two.
As you should be.
Gimp...
Ahh but the reverse is true. If you are the administrator of a small company, like I am and you run Windows XP on your clients, like I do and they get hit with some new worm that cost your business a couple of days of downtime, you have to be prepared to take the heat on that one.
We are in the process of evaluating switching some of our clients to Linux. I am not going to say it WILL happen, but it is being evaluated. I would assume that any company that is looking at switching would also go through a similar process. If they do a good job then most of the issues will be flushed out.
Are you talking about a Jtable? If so, I understand that it isn't easy to work with a Jtable or a Jtree. Both are somewhat complex. However, like all things that complexity comes with some power. If I remember the issue is that it uses a very strick model view controller pattern.
We could debate if Sun would have won or not...
.Net.
I am more curious what objects you had to extend in SWING that you don't in
Can you provide some examples? One I can think that use to exist was the inability to limit text input on a jtextfield. However, as you mentioned, you extend it one time and then you are done.
Good point. Sorry about that. I have been a bit jaded about the whole Microsoft thing because of their tatics inside companies that I have worked for. That is another story though...
.Net makes codeing better than JAVA? Let's not debate the GUI. I think we will agree that good GUI development is hard either way.
I apologize for any attack on you.
I still would like to know what inside of
You don't like SWING... you say it is because SUN made it cross platform. Weird, I worked with SWING a ton for two years, and found it ok. There were things I didn't like about it, but in general I liked it. The learning curve was a bit steep, but once done it wan't bad. I personally love the fact that if I write an application on Linux (and don't use an X,Y, layout manager) it will look and feel good on a Mac, NetWare, Solaris, and even my PDA. No it doesn't use native widgets, but because of that it doesn't behave weird on different platforms. Do I have some issues with SWING? Yep, but all in all it works fine.
.Net is a better desktop app environment. How? What specifically is better? I am not trying to be a jerk here, just to understand it. I would think tools such as webstart would be something you would like.
When is the last time I saw a XYZ app done as a web application? Off the top of my head I can think about turbo tax. But in general I agree with you. However, I also believe that will change over time. The "idea" of applets will come again. People are sick and tired of paying for software they take home, use once in a blue moon and then have it crash their system after an upgrade or worse yet require some other piece of software to be loaded first.
You say
A Java CPU... Why? The only issue I have with Java is memory usage, not CPU. On a ton of test JAVA has run neck and neck with C. That isn't too bad...
Open source Java - Yep that would be bad, Microsoft would grab it and change it/criple it and then put it on EVERY version of Windows from now on. Then people like yourself would say "We have to code to Microsoft's JVM... it is the defacto standard". I have been there and done that. I am so glad Microsoft is out of the Java business.
I agree with a lot you said, and again I am not trying to be a jerk here.
:-) Getting windows to scale and run well (threaded), takes some good design. I have found SWING to be able to handle most things well, granted there were some issue, but there were issues with VB.
:-)
.Net to run well on anything but Windows.
.NET. It isn't bad "if" you know what you are getting in for, but it will probably come back and haunt you in the future.
It is funny you mentioned SWING. I spent two years of my life developing SWING apps, and found the learning curve to be quite large, but it works great. I also didn't come from an object oriented background and that made learning tough. So would I say that SWING is better on Windows than Windows GDI? Nope, but would I say it is good enough for a very large percentage of applications? Yes. An issue I see with Windows development as the GUI goes is what version/service pack are you coding for. A lot of people still have a mix of 95,98,98SE,NT 4.0,2000 and XP. Yes there are similarities, but there are a ton of issues. This is were ideas like Java's WebStart start to shine. You as a developer have some control on what the client is running. As I mentioned some smaller shops can control what versions of clients they have, but it gets almost impossible at the larger shops (Dow Chemical being one execption, but their standards have caused them to delay upgrades for years). So yes their developers have a standard, but that standard is old. If it isn't old, then it will be in a year...
Now, good gui development isn't easy, well I should say it isn't easy for me
So yes 99% of the people do have a Windows box at work. I agree. Will you agree that of that 99% no more than 25% are running the same version of Windows? I won't even count the numerous service packs that make apps behave differently. (Like most even numbered service packs
Now for SQL server V.S. Oracle. That was what the company wanted to use. Oracle isn't cheap, but neither is SQL server. Also, I just hired a good Oracle DBA. They are the SAME cost as a good SQL server DBA. (We use to have SQL server) (I live in Indiana, labor is very very cheap at the moment, we are a manufacturing state and have been hit hard by the recent economy). I also have worked with DB2 and I agree with what you said about not discounting any DB. This includes of course PostgreSQL and MySQL.
Lastly, I think you and I agree on a lot, I just want to make a few points.
1. Microsoft will NEVER NEVER allow
2. If/when Microsoft looses marketshare that code will become a liability if it needs to be ported.
3. Competition is great. It makes everyone better.
Ok, since this is getting long, I add one more point; and pick on Oracle. Oracle has a development package called Jdeveloper. It can use an object to relational mapping tool called BC4J. Once you have your BC4J objects you can call them via their front end tools (UIX controls). This looks great on the surface. I was able to bang out an applicaton in 50% of the time it took previously. Like you I do a traditional development style. This tool looked to be the saving grace, and our estimates for projects looked like they could drop by a very large margin. Ahh but then the issue came in. I won't go in to the HUGE list of issue, but suffice to say we will not be using their stuff for development for a long while. This is exactly how I feel about
Either way good luck.
If you worked with JSP then you know of things like custom tags. In my opinion JSP is quite a bit ahead of ASP.
.net stuff use anything but Microsoft. The same isn't said for C or Java.
.net then you need to seriously look at Java again. It does everthing, and then some(container managed entity beans), that .net does. The core difference is that it doesn't lock you in to Microsoft.
.Net will never work well on any platform other than Windows. Microsoft will make sure of that.
You like Javascript the best? Wow you are the only person I have talked/written to that says that. Weird. Most people hate it.
If you believe that Mono will ever bring quality C sharp apps to Linux then you better start learning another language. Name an instance in Microsofts past that shows how they have worked with another vendor to run their apps on a non Microsoft platform. I am glad that these people want to develop Mono, but to be honest I can never see any I.T. shop that is doing
Again, if you like
I am not trying to be a jerk here. I just want to warn you that having worked with Microsoft for a while, they will make things easy at the cost of security and scale. They will NEVER allow their cash cow Windows to be jeprodised. Their apps tend not to scale well. If you want I can point you to some shops that are probably going to close soon that bet the farm with Microsoft. One in particular has had the boys from Redmond in for a while and this was some of the recommendations.
1. Reboot your servers every night. Granted this company has over 400 Win2k boxes.
2. Buy more hardware. Yet people like me have shown that you could scale to what they want with some effort using non Microsoft.
The last one kills me.
3. Get rid of Oracle and go with SQL server. Yet their DB and connection is not a factor at all. This is the ONLY non Microsoft product they own, and yet Microsoft "techs" want it ripped out.
I can name many other companies if you like. I don't want to say Microsoft won't work at all, it will, and in some smaller shops it might work well. I just want to warn people what they are getting in to before they spend their hard earned money.
Please Please Please take from this that