Linux has recently (July ’98) been getting quite a bit of publicity in the press.
The focus of these press reports is the growing groundswell of support for Linux use in corporations. There are huge companies that cater to huge corporate clients who are now bringing out products that will run on Linux. Examples are AT&T ( Inferno ), Informix ( INFORMIX-SE ), and Oracle ( Oracle 8 ).
I have also waded through a number of flame wars on the newsgroups and mailing-lists. The theme there has been: ‘ It’s a good thing that Corporate America is waking up to Linux’.
However, one of the points in the flames wars, that I happen to disagree with is the question of support.
People often claim that one can get better support within the Linux community (via newsgroup and mailing-lists) than one can get through the official support lines of Microsoft Corp. Certainly, I have found the support lines of Microsoft Corp. to be a very frustrating experience. But, on the other hand, the newsgroup, and mailing-list support also has major drawbacks.
A case in point. I recently added a third hard-disk to my machine at home. When I booted up my Linux OS, I found that it had stopped recognizing my cd-rom drive. After twiddling around with it without much success, I took this problem to the newsgroups.
I did receive replies from people. Some of them were helpful. Some were not. The net result, however, was that I did not succeed in solving my problem. I then took it to the mailing lists. After a couple of days, someone sent in a bit of info which enabled me to fix my problem.
What is wrong with this scenario?
To me, as a hobbyist – nothing. I am glad that I could solve my problem. If I were a corporation seeking a solution, however, I can see two problems. Let’s deal with the minor one first :
1. The Signal to Noise ratio: Lots of people contributed to solving my problem. However, to a corporation, the more invalid replies there are, the less confidence they will have that newsgroups / mailing-lists are a way to solve the problem. Sooner or later, their confidence will drop so low that they might resort to drastic measures – like switching to a commercially supported product.
To this, I can almost hear the wail of responses saying ‘But you have the source code! You can debug it’. The problem here is that IT is not the primary business of most corporations. If it were, then yes, they could afford to hire hackers to debug code. The programmers in most IT departments are hired to support applications related to the business, rather than tools / OS support.
2. Accountability: I think, most people have the wrong reasons why Corporate America has a liking for accountability in the support arena. The reason that I have most often heard goes something like this: ‘If something goes wrong, the corporation would like to know that there is someone that they can sue’.
While there may a grain of truth in this, I do not think that this is the chief reason for such a requirement.
Let’s take the case of my cd-rom drive. If I were in a corporation where time is money, my objective would be to get my problem resolved as soon as possible. I would like to be able to call up someone and say ‘Hey, this is a piece of code that you support. I need this problem resolved as soon as possible’.
If I were depending on newsgroups / mailing-lists for my support, this is one statement that I absolutely could not make. Even if I communicate directly with the developer/maintainer of an application, it is quite well known that most of them are so deluged in the mail that it would take them a significant amount of time to get back to me.
I once worked at a moderately large insurance company on the east coast. One day, their mainframe crashed and was down for about 15 hours (a very unusual thing in the mainframe world). The corporation claimed to have lost more than a million dollars in revenue and lost business because of this downtime.
Certainly, if time were that expensive, businesses would not like to use newgroups / mailing-lists as a line of support.
Yes, there are companies out there today who offer support for free software. And I feel that this is the right way to go for corporate customers. But there is no guarantee that the support lines of these companies will not have the same problems that plague the support lines of other less-liked OS suppliers.
Let us take a closer look at companies which provide such support. I recently had a chance to interact with a very prominent vendor of Linux over a question of support. The point of contention was token-ring card support.
Yes, token-ring technology is all but dead. However, I am currently working at IBM, and the IBM intranets are built with token-ring technology. There were several old, down and out machines I came across, and I decided to try and run Linux on some of them.
Token-ring cards are supported in Linux. But not very well. For example, token-ring cards that use DMA instead of shared memory are not supported. There is no support for PCI token-ring cards made by IBM, though there is a vendor called Olicom who sells PCI token-ring cards and also have a driver for Linux(!). I did my research and found all of this out, determined that the machine I had access to would run Linux before I bought the Linux distribution from the aforementioned vendor.
So I installed Linux on the machine. The installation went fine, but the machine refused to network. I contacted the vendor support by email and informed them of the problem. A day later, they got back to me saying that, er.., it’s possible that the card was not supported. I wrote back saying, that I had done my homework and knew that the card was supported, so would you please help me with the install?
I eventually discovered that there were some problems with the way Linux autodetected the card that prevented it from working on my machine. So I got a newer model of the token-ring card and installed it. It took me a while to figure out how to get it working, but I did it eventually. It took me about 4 business days for this.
Oh, and what happened to the vendor? They got back to me after three business days and told me that a more recent kernel might help.
Is this good support? An individual user who is determined to stick with Linux might think so. Certainly, a businessman with a business to run wouldn’t think so. What are the problems with this kind of support? Let me explain.
1. It took too long. Four business days to fix a machine dead in the water? That’s too long by business standards.
2. The hardware support policy was inconsistent. The vendor had stated that there was some hardware they would fully support, and some they would do only a half-hearted attempt at supporting. (Those weren’t the words they used – it is my translation of what they said. They call it tier-one supported hardware and tier-two supported hardware). To survive in the enterprise, by that I mean the GE’s and the IBM’s and the Walmart’s of today, you have to have a policy that is more consistent. One that either says that you will or you won’t support a piece of equipment. Guess what? Token-ring cards are tier-two supported hardware.
3. They didn’t solve it. I did, and no thanks to them. This may not seem a big deal to an individual running Linux. But to a corporation, this leads to loss of confidence in the vendor. And after that, the writing is on the wall.