Expriences with Open Source Web Technologies?

_ZorKa_ asks: "I have been asked to give a presentation at an up and coming technology expo by my employer about 'Open Source Web Technologies'. Having taught Linux courses and open source web languages at a local college I feel fairly comfortable in speaking on the subject in front of a large crowd. So, instead of giving some of the same old boring open source stories I normally use I thought I would poll the Slashdot audience and see if there is anything of particular interest I should mention during my presentation. Maybe you have a war story to tell, or some advice to give. Anyone have any great stories that I could pass on as to why you *should* use open source web technologies in a business? Anyone have any links to some sound numbers that can back an argument up with. Even if you have reasons why a business shouldn't use open source web technologies I would like to consider those as well. I'm all ears!"

Comments

Apache; BRL

[brlewis]

Without getting into whether or not this is a good thing, large businesses want to use what everybody else is using. It is generally assumed that everybody is using Microsoft. You should correct this misconception by citing the netcraft survey.

Please do mention that while the more prevalent free server-side web technologies are scripting languages designed for people with a Unix/C programming background, they aren't the only options. My Beautiful Report Language is easy to pick up for anyone who knows HTML and SQL. The syntax is conceptually similar to HTML/XML, but less verbose and easy to visually distinguish from markup.

MySQL + PHP Story

[Anonymous Coward]

I work for a web development company in Draper, Utah. Initially when I signed on we were running on a purely NT atmosphere with ASP, JSP and Coldfusion installed. All of this was running on the backend MsSQL server hosting up a ton of registration data for companies. The pain in the ass part was the up time for the servers. They had hired a MCSE that managed the NT boxes, they went down (i kid not) atleast 10 times a day. The people at our colocation service gave us their ICQ numbers so we can just icq them when the server needed to be reboot rather than having to go through the whole rigermoral of calling through the tech support lines. We decided to move over to Debian Linux stable. After a very painful transistion (mostly to the lack of knowledge on how to properly export databases out of MsSQL from our MCSE), we have a fully running apache,php + mysql implimentation (with JSP support, Coldfusion support and even ASP support thanks to ChiliSoft). This is increadibly useful in that our servers have not gone down in over 2 years that i've been here. We also were able to get rid of the annoying MCSE, seeing as how we could hire a consultant to come in once a month and double check that everything was all right. Moving to linux was the best choice this company ever made, and I haven't even gotten into the whole money savings thing. MySQL + PHP also allows you to integrate your software into other open source applications such as ghostscript or the pdf libraries. We were able to dynamically create certificates made from PDF's with peoples signatures and names on them for clients. This is something that can be done in windows but requires you to go through there Printer OLE Com functions, not to mention it halts the system while the print job is finishing (oh and costs 5000 dollars for the ASP plugins.).

That's my beef with Windows NT, etc, doesn't run what you want fast enough and costs a huge amount. Not to mention it crashes 10 times a day :-)

wartime stories

[Roadmaster]

This company organized a huge media-related event, and they wanted to have an on-line registration system. Initially they called in a team which developed the system on Windows NT servers, with ASP and SQL Server. They took 3 months and when they finally delivered the system, it simply crashed under the heavy load.

That's when they called us in. This was 2 days prior to the registration day, so we were in a big rush. We, of course, advocated open/free technologies; we suggested Linux, MySQL, apache and perl. They said OK, so we started working. At the same time, the NT team worked on their server, tweaking it to see if they could withstand the load.

30 hours later, and on a server that was about 1/2 the capacity of the NT team's server, our system went on-line. It turns out it, too, crashed under the load. Heh. But while the NT team was busy rebooting their server just once, we installed Linux on 4 additional computers (desk-grade IBM Aptivas) and ended up with a round-robin configuration, with 4 web servers and a backend database server. Our system handled the load so well, that this time the company' firewall crashed and they had to update it.

I still like to use it as an example of how free technologies and tools enabled us to re-engineer 3 months worth of work from the NT team, in about 36 hours.

my stories...

[spike666]

i have 2. the first, my department wanted to install web servers onto our many distributed NT servers. our installation group said "no, we dont want to support IIS cuz its buggy" our technical architecture team said "well, cant you just do the same thing with file:// urls? or write some other program to do it? (my answer to that was that i'd just install apache and say that i wrote a server based on industry standards..." i showed my manager that we could easily install Apache win32 and go from there, but he chose to push the IIS issue and finally that got installed.
months later, when we had buttloads of issues with IIS, he came back to me and said that in retrospect it would have been easier to just install apache...

the second story - i was allowed to go ahead and implement Apache Tomcat for a subsequent project, in order to fufill a java based set of core components. it works great, and i even have it hooked into our IIS servers (still have those buggers) and everything works great. the big problem now is that on our 60 person team, theres only 5 people who know java, and management doesnt want to continue with it, even thought our corporate direction seems to be headed towards java!!!

ok, in retrospect these probably arent the right stories to be regaling the populous with, but they're what i got.

businesses dont care about stability...

[spike666]

one thing that i came across at my company, which is very microsoft flavored (but moving away slowly) is that in terms of stability, the management has become so used to the lack of stability from Microsoft products that they scale up the hardware to adjust - aka making sure they have abundant backup servers that are load balanced or hot swapped in when problems occur.

its quite sad. even the need to bulk up hardware in memory , cpu and disk does not phase them. - one example where we could save millions in licensing and hardware upgrades by moving to a smaller, cheaper, faster linux solution was thrown out without consideration because "well we should use microsoft right?" and my favorite quote "well i'm sure Bill knows what he's doing."
i think in the future i'll counter with "well linus knows what he's doing, and he's much cooler"

Open Source Anecdote

[Anonymous Coward]

You can't more open source than this.

Show them the money

[mckwant]

I'm a UNIX admin, but I just got my MBA, and the following is what I'd be interested in, if I were attending such a seminar.

1) Switching Costs. The cost of using OSS != hardware costs, it's also the time and effort spent moving over to the new system. How much is it going to cost me to switch over from my current platform? Mention ChiliSoft's ASP stuff, the fact that you CAN run front page extensions on apache (not that I'd recommend it, having installed it, and, man, it's a biyatch), and that you can train sysadmins by using an old desktop PC.

2) Recurring Costs. How much can OSS save me, either through being able to effectively ignore the Windows licensing mambo, increased uptime, or decreased administration costs. Also somewhat important is that you can run the exact same software on a lesser development box if you want, and the marginal cost of that second box is negligible.

3) What happens when we have a problem that my sysadmin can't solve? This is probably the largest barrier to OSS in the private sector, so I'd suggest you deal with it head on. Of course, I don't really have a good answer for that. I haven't found any CIO level person willing to bet their livelihood on "well, there's a really active newsgroup" type arguments. You might also point out that training is available for Linux at comparable cost to NT, so that's pretty much a wash. You might also mention the virtual worthlessness of the MCSE as a gauge for finding a valued sysadmin.

4) Can the company find staff? Finding UNIX geeks isn't simple, finding good ones is downright difficult. I think this has become simpler with the rise of OSS, but it would be a question I'd have if I were a boss. Mention geekfinder, and maybe some other OSS-oriented employment sites.

What I would NOT do is mention that Apache has x% of the web server market. This is contrary to an earlier post, but that Netcraft survey can be invalidated by a quick "yeah, but a bunch of those web sites are academic, or college kids, or hobbyists, and aren't actually used to conduct commerce" comment, which is somewhat valid.

I would also steer clear of uptime and security claims. These guys aren't likely to be on the front lines of keeping those boxes up, and won't appreciate the pain of keeping NT going.

I would mention the cost savings of OSS not as an initial cost, but as an ongoing cost, basically because a) these guys probably have a web site going already, so they'll have to switch over, and b) even if they are starting from scratch, they'll have a budget to work with, so while cost is an object, it's not the primary one when approaching the purchasing decision.

I hope it goes well.

Re:Show them the money

[_ZorKa_]

>What happens when we have a problem that my sysadmin can't solve?

I live in Michigan (outside of Detriot) where there are tons of consulting companies who specialize in providing these types of services for companies. One that comes to mind is located in Ann Arbor, MI called "The Linux Box" which is a firm that specializes in nothing but helping companies use and maintain Linux in their corporate environments.

I am from Mississippi originally, and you are right in that finding a Unix geek there would be extremely difficult unless you are in a bigger city (which there aren't that many).

Something I like to ask IT managers is, "The last time you had a problem with an NT server did you solve it by calling Microsoft?" I have probably spoken to hundreds of people in this type of position and I haven't gotten a YES yet. Why? The main reason is that Microsoft doesn't provide support directly for its OS. I sold hardware for about 5 years and everytime we sold a PC, we (being the manufacturer) were required to support the operating system and whatever we sold them. Sometimes the on-site firm we used as support had to go out and try to fix it. If we called Microsoft for support we never got an aswer or a bug fix. We ALWAYS had to resort to forums and online news posting to solve the problems, or contact the manufacturer of the products we used.

I guess my point is that Microsoft has done a great job at convincing the IT managers they have support, but the truth is they don't. It is another illusion Microsoft has everyone buying into.

I do mention the numbers about apache and others though. As someone else pointed out "businesses use what everyone else is using" and this is a key point to selling it to non-technical people who really don't know any better. Their mentality is "If it is good enough for the goose, it is good enough for the gander".

I met with the executive vice president of a very large advertising agency this past week about using our companies product which has been built using open source technologies. During our introductions he kept mentioning Windows and ASP pages. I didn't really know where he stood as far as using a product built to run o pen source technologies. So I asked him, "I am going to mention two words to you and you tell me what comes to mind........Open Source".

He thought a minute and said, "Well, stability, scalability, more functional, and ease of use". I don't know where he got the ease of use but everything else he said I couldn't believe.

Re:Show them the money

[spike666]

i work for a large corporation that uses Microsoft (go figure) and we do call upon MS for support alla time. especially since we dont have the time to pick though the BSOD dumps, and we pay them for it. but you are very right - they dont give us much back in the way of real support - other than to pull up MSDN articles we could have searched for ourselves. and usually we find the real answer ourselves in 1/2 the time.
BUT. the problem is that the management perceives that microsoft is supporting us. it isnt about the REALITY of support, its about the PERCEPTION, and more importantly, the BLAME of support - management wants to be able to tell their boss / the clients "we think its a problem in the Microsoft stuff, Microsoft is looking into it" i've consistently found that its all about CYA - Cover(ing) Your Ass - as long as you can blame a vendor, its ok. thats the real reason they're scared of using OSS. you cant blame a website, you cant point to a newsgroup and say "they're looking into it"

by the way, you said this vp kept talkin about windows and asp pages, but he had the right answers about Open Source - was he just a talking head or did he know what was what?

oh, and beware of the management using your answers to build up a case for OSS/Linux and then taking that as leverage to Redmond in order to cut the licensing costs...

Re:Show them the money

[_ZorKa_]

After meeting with him for 4 hours I got to know him pretty well and he is a very informed consumer. He was mentioning windows and ASP because that is what they have used in the past for the web sites they build. It just turns out that NO ONE ELSE had ever approached him about using open source technologies. He knew about them, but none of the vendors they were using offered open source technologies.

I think this really drives home the point that Eric Raymond makes in his speaking about going to the "top" to get the company to use open source. If you get the decision makers on your side then things will start to move.

If more companies like ours embraced open source to build its products with and support the product on open source platforms it covers two of the biggest objections, cost and support. Our cost is lowered since we are using open source products to build it with, therefore there are no license fees to pay. And two, we support it very well on linux. If the product works and works well, will save your business money after you buy it and is supported there isn't much of an argument against using it. If the partner your company is working with doesn't use open source then you are going to wind up using windows and paying a lot for it.

I will say this, the open source community needs a few things to kick this thing in the rear:

1. Business tools - like ACT, goldmine, contact management, accounting packages, inventory management tools etc. The problem is that "geeks" don't won't to write stuff like this so we leave the job to closed source software companies to develop these apps and make millions. Sure we almost have evolution but how long did it take the open source community to say, you know this Outlook thing is pretty cool and companies are using it.

2. Kill off the stupid forking of open source projects. What I mean by this is if you go to sourceforge.net there are over 29,000 projects. Most of which are things that people are duplicating over and over and over and over. The programmers that are cauusing this say how bad a project is and then go off and start some lame project that never goes anywhere. These programmers aren't doing the open source community any good. They should FIRST AND FORMOST join an existing project and make it better. This would make the SuSE distribution a lot smaller (hahaha).

3. Geeks need to become salesmen and start speaking in terms a business or another person can understand. Telling your boss that you are "31337" and he should use linux because your uptime is 126 days is not convincing enough. I have found that business owners respond positively if given the proper pitch.

Re:Show them the money

[mgkimsal2]

What I would NOT do is mention that Apache has x% of the web server market. This is contrary to an earlier post, but that Netcraft survey can be invalidated by a quick "yeah, but a bunch of those web sites are academic, or college kids, or hobbyists, and aren't actually used to conduct commerce" comment, which is somewhat valid.

Two points on this:

1. If you look at the 'active' sites aspect of the netcraft stuff, http://www.netcraft.com/survey/index-200007.html#a ctive , you'll see that Linux and MS are pretty much tied, kinda laying to rest the "registrars just having 100,000 domain names on one server" accusations.

2. The "commerce" part - measured generally by what SSL servers are running, is somewhat skewed. I know I've seen a netcraft things before, but can't find it. http://www.netcraft.com/surveys/analysis/https/200 1/Jan/CMatch/oscnt_all.html is the closest I can find. Linux with 18% of SSL server's isn't bad, anyway. But more to the point, my recollection is that verisign, the only viable SSL certificate registrar for a long time, wouldn't issue certificates for Apache until some time in late 1998. Perhaps I'm wrong - someone please correct me on this if I am. If it IS true, factoring in inertia, it's surprising Linux has as much as it has, if NT had 2 years 'lead time' with SSL certificates.

Re: uptime and security - depending on who's listening, they may not feel the pain of getting up at 2am due to downtime, viruses, etc., but they see the bill for that at the end of the month. They still notice the imapct where it hits - at the bottom line. If someone could reduce their support costs 30% a month because of less server maintenance, they'd listen.

Re:Show them the money

[mckwant]

I suspect that the guys we're talking about will follow the path most often traveled by guys in similar situations to what they're in. So the 18% you quote is a lot more accurate than the 45-50% I recall seeing on netcraft for web servers in general. I'm not arguing that Linux hasn't come a loooong way, nor that you can't do it on Linux, since you certianly can, but if you're using the "they'll follow the crowd" argument, that argument doesn't go through Linuxville. Sorry.

Re:My experiences

[_ZorKa_]

I don't really count this as a viable argument as this can swing to both sides of the fence. Usually when something like this happens it is a case of YOUR COMPANY didn't hire the right people who knew what they were doing. It doesn't really matter if we are talking about Linux or Windows in this case as both operating systems have fallen prey to this type of "blame" (which they are going to loose anyway cause stuff isn't working).

I can see it now, someone at your firm "wanted" to be a Linux savior but didn't really have any idea what he was doing, but he knew that Linux would save the company money and would meet the needs of the company. So this guy/gal recommends Linux, people buy into it and when it comes time to tweak it and do a corporate setup the guy/gal doesn't have a clue as to what they are doing. Of course, the person isn't going to say "I have no idea what I am doing, we should call in a Linux consulting company for some help" in fear that he may loose his job.

Some people live happy with one solution or the other and stories like this could have been solved if the right person was doing the right job. I would sue your human resources department for not knowing enough about technology and how to interview people. Company's only survive if the people they put in place KNOW WHAT THE HELL THEY ARE DOING.

"Somewhat valid" objection to netcraft

[brlewis]

It's easy to point to a survey and make "somewhat valid" objections to the results. Survey numbers are generally not scientific measurements. The right question to ask is, what do these numbers tell you?

1. Non-MSFT web servers are not an obscure corner of the market.
2. Just because MSFT monopolizes many users' desktops, doesn't mean it monopolizes all other markets.
3. If you look at what software is running on major web sites (e.g. yahoo, google, CNN, nytimes) it usually isn't MSFT. (This is a useful learning for people even if the non-MSFT software is proprietary.)

In summary, a lot of people need to learn that choosing something other than MSFT on a web server does not put them outside the mainstream.

Re:My experiences

[Anonymous Coward]

Been there. Done that...

At the job interview: What do you mean you want $60/hour? I can hire a high school student for $8/hour! Get the hell out of here...

A year later, when they call me back in desperation: What do you mean your prices have gone up?