Summary: As Aurora looms, the new OS/2 Warp Server for e.business, I find myself reflecting on where the industry and Warp Server are. I'd like to share these thoughts, facts, and opinions with you and hope you will share yours with me.
I was recently asked (by an IBMer) ... "Why are people complaining so much about NT?" My response is: Because familiarity breeds contempt. People are now using NT and becoming more familiar with what it can do and how well it does it; not always living up to expectations. Familiarity breeds contempt.
We might add resentment due to the fact that many feel they are forced to use NT. The "market mindshare" leaves them no choice. People tend to resent a lack of choice. The issue comes more serious if the product they feel "forced" to use simply doesn't do the job to their satisfaction. And the promises of things getting fixed in the next release, service pack, whatever, are getting thin. The big fear isn't that Windows2000 won't fix existing problems, but that it is sure to deliver a entirely new set of problems. That would stand to reason for any new release of a large software product, even if the track record wasn't already there.
Computers are supposed to be reliable. But this is Windows we are talking about. For some reason, which has always eluded me, users have been quite accepting of Windows' frailties. A satiric item was circulated years ago called: "If Microsoft Made Cars". It was a jewel. My favorite line (paraphrased from memory) was "Now and then, for no apparent reason, your car will stop working and you will be returned to your trip's point of origin. But, for some reason, this is OK with you."
More recently there has been this hoo-haa about the Windows 49.7 day bug. It states that some machines, which run for 49.7 days, may experience problems such as a lockup due to a timer that counts every second and eventually counts higher than the memory allocated for it can handle. The user response has been along the lines of "Whose machine has stayed up long enough to detect this failure?" According to the press item, most Windows users would be happy with a system that stayed up long enough to have the problem. But this level of inexplicable tolerance does not generally spill over from desktops to servers. Or does it?
Window's frailties might encourage businesses to try OS/2 but they have to contend with IBM's ambivalence towards the product. If the company selling sends a message that they don't want people to buy it, any sales qualify as demonstration of miraculous faith and indomitable courage. Or just a damn good product. We've all see the comments by IBM execs in the trade press, and comments about not wanting to "grow the market", misplaced pride in stating which markets IBM chooses to serve. And my fear that IBM NCSD is about do this again by backing away from an "Aurora client". IBM is mayor of Mixed Message City.
As IBM Personal Systems Group was posting their one billion dollar loss, International Data Corp (IDC) was releasing a study showing that shipments for Intel server sales jumped 15% in 4Q98. And this happened despite a 4% drop in overall server revenues. Message: more and more folks are moving to Intel servers and this growth of Intel based servers favors NT because this space has been largely ceded to NT, even by IBM. Here I'm referring to IBM PC servers company and the IBM Software Group, including Lotus, which has demonstrated more support for Windows NT Server than for OS/2 Warp Server. IBM NCSD (formerly PSP) takes Warp Server's opportunity seriously. But if a tree falls in the forest ...
This same IDC study said NT was doing well because of ease of use and familiarity with WindowsNT .. the fact that NT is very available from vendors, and the application support .. they could have added device support. They may have, I don't have access to the entire study. But the study goes on to say that this success continues despite the need to overcome "what some call the inherent unreliability of NT, which many users claim results in 20% downtime." I just don't understand how a business can expect to survive if the servers are this unreliable, no mater how inexpensive the hardware is, no matter how available the applications may be.
What's sauce for Linux should be gravy for Warp Server
Several months ago I crawled out on a limb and said that Windows NT had hit the high water mark. It will be down hill from here. This was the end of the year. I was recently validated, to a limited extent, by industry figures about Linux. Linux had out paced NT Server in new installs in 1998 according to IDC. 1Q99 Linux installs should be very high, perhaps equal to all of 1998. Of course, IBM is quickly zigging over to Linux as The Next Big Thing. Some wags say that it was IBM's software business plan to make NT "strategic" which put the jinx on the product. If so, the same fate will befall Java and Linux. I don't think so, though I think it's a good joke.
The problem with NT, which people are beginning to understand, is that it is brutally oversold. It simply can't get all the jobs done. It suffers, in many environments, from reliability and performance issues ... and it is mighty expensive. Some folks talk about running web servers on NT and spending $75,000 for all the software and hardware. That's a lot to spend and then find you have Y2K, security, and reliability issues. Severe issues of deploying software, changes, upgrades, service packs, fixes, and so forth. And users are told that Windows2000, mother of all software, will fix that. As Count Floyd would say, "Chilly scary theater, children".
Part of the irony here is The Tale of Two Servers. On the one hand, you have NT. A reasonable product which would appear designed and best suited to replace Novell servers and "small" Unix boxes. But NT gets positioned as the all everything system. Oversold and placed in environments that are beyond its current capabilities. But, it is MS's only server today. And when the only thing you have is a hammer, everything looks like a nail.
IBM has a full tool set of servers, and seemed to be ready to market this way. But recanted. Now, I can't really tell you what IBM's server story is. IBM has a full range of excellent servers, hardware and software, from OS/2 Entry to water cooled monsters. But you would think they are selling NT ... and soon, Linux. Microsoft oversells it's one product. Then IBM comes along and oversells Microsoft's one product. Clue for IBM! For the rapidly growing Intel server space, there isn't a better, more capable, more reliable, price performer "out-of-the box" (as opposed to Linux) server than Warp Server SMP.
The widespread use of NT can raise awareness and appreciation of Unix, Linux, and, I hope, OS/2. I recently spoke with a representative of a national services and consulting firm who told me that they are moving folks off OS/2 and on to WindowsNT. He didn't like doing it, but "It's what the customer wants." But my hat is off to the consultant in Texas who said, "When they ask me to put them on NT, I tell them I'm not Dr. Kevorkian".
Stay with, or move with all due haste to, Warp Server
Forgive my obscure references, but when Steve Martin was really, really big .. he made a movie called "The Jerk". In that movie, as he sets off from home to find his place in the world, his father takes him to the side of the farmhouse and points down at something on the ground. He says a four letter word. Then he points to a small can of shoe polish in his hand and says, "Shinola". He repeats this important message, pointing to the ground, then to his hand, hoping the ability to tell the difference will assure his son's success.
Well, seeing folks leap like lemmings off OS/2 Warp Server makes me think many fathers did not impart that wisdom to their children. There is no good reason I can think of to move off OS/2 Warp Server. At a relatively recent training session, held by IBM in Austin, some personnel from an large airplane manufacturer in the northwest told this story: There are forms on doors which require someone to sign and date the form each time the door is unlocked. There was a door which, based on information on the form, had not been opened for several years. No one could recall what was in the room. So, they unlocked the door and there was a machine running LAN Server ... still running, still part of the network, supporting users in an uninterrupted fashion for years. Reliability. Availability.
One large Lotus Notes user has reported to IBM that Notes Servers on NT are providing 75% availability. The Notes Servers on OS/2 Warp Server are 99+% available. What could be causing this? Is the Notes code for Windows so inferior to the Notes code for OS/2? I'll be kind and say, that's unlikely ... certainly would not be intentional. Lotus, like most parts of IBM writing software for the Intel space, is consumed by a desire to run on NT. Could OS/2 Warp Server be that superior to NT Server in a Notes environment? I'm thinking "yes" but it would probably kill Lotus to admit that, and maybe I'm wrong. I'll be generous and say "I have no idea".
I remember the excellent work that Austin PSP did on Lotus Notes performance benchmarks. Sam Emrick and his team ran the benchmarks and made suggestions for performance and tuning. I recall Sam supporting the workload of one thousand Notes users on one Compaq server. Imagine the effect 75% availability would have in that situation. How many more NT Servers would you need to equal the load carried by the Warp Server systems?
Last month I spoke with an IT technician with a world wide insurance firm, working out of the Atlanta site. They are in the second year of a one year migration from Warp Server to NT. They have found they need to replace each Warp Server with two NT servers, which costs three to four times as much ... but, for some reason, this is OK with them. Same story from a consultant in Raleigh. Second year of a one year migration and they are over budget and only 40% complete. But what choice to they have?
I think we have identified a trend here. Folks seem to be banging their head and saying "It hurts when I do this." When we respond: "Stop doing that!" they tell us they have no choice. But there actually is a choice. OS/2 Warp Server offers a choice. And I think some companies may be at the point of realizing that they don't have to settle for NT. IBM's launch campaign for Aurora ought to be "Now, you have a choice." But there really is no need to wait for Aurora. OS/2 Warp Server has been here for years.
OK, so there's reliability, availability, and performance. Anything else?
Of course! But shouldn't that be enough? Isn't that the point and the value of the server? Well, OK. OK. There is also support for the network. Let's take a moment to consider what Warp Server provides:
And something else which is not immediately apparent. Support for Warp Server. Despite the "everything, including the kitchen sink" approach to Windows2000, Warp Server is an extremely complete NOS and deserves some recognition. This is something else that get's better with Aurora, but even today's product offers excellent features and benefits from some superior industry support which builds on Warp Server:
OS/2 Warp, the desktop client, gets slammed a lot over issues of applications. This is less of an issue when servers are discussed. But it's less true for device support. Even servers need to stay current when it comes to device support and that costs money. IBM has not shown an inclination to invest much in Warp. But even given the fact that the selections may be limited, I believe the overall value proposition on Warp Server is still superior. How does device support help a server that can't stay up?
I have often listened to analogies, similes of Warp and comparisons with Windows. Warp as the Checker Cab, a specialized and industrial solution designed to do the best job, low maintenance, high reliability. Just well designed and effective, if not pretty. With Windows we get the VHS - BetaMax story. Not a bad one.
Most folks at home don't bother with beta ... but when you go to the video department of a media house, or TV studio ... it's all Beta. And that may be what pulls OS/2 back from the brink. The requirement for something that is industrial strength and works. The fact that is less expensive hardly seems to be factor ... which I've never understood.
|Copyright © 1999 - Falcon Networking||ISSN 1203-5696||April 1, 1999|