Sun = the RIAA of the computer industry?

38 thoughts on “Sun = the RIAA of the computer industry?”

1. A better idea? Don’t try so hard. Many major corporations still love Sun boxes. Heck, I would prefer a Sun box with Solaris (and enough GNU tools to make it usable) over a Lintel box any day.

   The problem is, Sun hired so many people to write exotic software and provides services. They are extremely proud that this software (Java) runs on many different platforms, and they give away this software for free. That doesn’t make sense. It doesn’t make anybody buy Sun boxes and they just waste money.

   There is a market for massive multi-processor machines with big storage. While geeks and scientists like cheap distributed systems, banks and many other big companies don’t. They want big boxes they can rely on. Sun is, and can remain, a major player in this market. All they have to provide is the hardware and OS plus some basic services. Oracle, Sybase, SAP at al will keep providing their software to keep the business buying Sun.

   Sun needs to realise they are no longer the dot in dot-com and that the hole dot was a moot point for their core clients to begin with.

2. Sun will always make more money selling to businesses than to individuals. RIM has the same strategy with Blackberry.

   Sun’s recent strategy seems centered around becoming known as “The Java Company”, which, while not incompatible, is a shift in focus from “The Network Is The Computer”. While they are still giving away Java for free, it seems now they want to compete with BEA in addition to IBM and Microsoft and offer an entire suite of Java software rather than the SDK and reference implementations.

   This is probably a smart move. While Bas is right that there remains a strong market for solid dependable Sun boxes, that business alone does not provide a basis for Sun’s current valuation. Sun would have to scale back considerably and tell Wall Street that they won’t be growing much anymore, but they hope to continue building and supporting nice reliable servers and maintaining share in an increasingly competitive market.

   Sun seems to be trying to become more like IBM, having strengths in both hardware and software, but with a greater focus on Java rather than Linux. IBM has WebSphere, but seems to be focusing their marketing on Linux. I guess Sun can compete with IBM if it turns out that companies are more interested in building Java applications rather than native Linux applications.

   Sun should probably buy some consulting services to compete better with IBM, rather than getting into the cable business.

3. Cut the the MS Tax.

   Push forward, *hard*, with a desktop OS that’s cheaper than Windows, but is completely interoperable with Windows. In other words, do exactly what they’re doing now (or trying to anyway). It’ll be tough considering how much of Windows is closed off, but that seems to be their best bet.

4. Ben: How can someone make money being “The Java Company” or “The [some computer language] Company”? The market for programming tools and execution environments is pretty small. I don’t think IBM ever got huge profits from the imperative languages that it developed, e.g., Fortran and PL/I. Particularly now that a Java-style execution environment is bundled for free with Windows (and comes with C#, a language with significant improvements over Java though still lacking many features from Lisp, ML, Haskell, etc.), and a free open-source clone of .NET has been developed for Linux (http://www.go-mono.com/) it seems unlikely that huge profits could be made in this area.

5. In the financial services industry, companies are ripping proprietary UNIX boxes as fast as they can, and replacing them with GNU/Linux on Intel boxes that are 1/10th the price. Are they as good? Heck no. Are they good enough? Yes.

   Some server applications still require large multiprocessor boxes, where Intel boxes won’t cut it, but in this market Sun has to compete with IBM and HP.

   Unless they figure out a way to differentiate the hardware, I don’t see a bright future for Sun. Software sales, which are miniscule for a hardware company like Sun, aren’t going to cut it.

   It seems that larger computers eventually get their market taken away by smaller boxes, as the smaller boxes become more powerful. Witness UNIX boxes taking over the minicomputer market, or PCs now invading the UNIX market. Mainframes have been the one exception to this rule.

6. How about: make big contribution to Bush reelection campaign, get big contract setting up web servers in Iraq. That would keep them busy for a while.

7. Well, if they want to ride the Java pony into the dirt *and* stay in the hardware business, maybe Sun could build a native Java machine — perhaps based on one of the Transmeta cores. Then commission a Linux port.

   Such a machine would presumably run Java applications pretty well. Write once, run anywhere, run better on a native engine y’know. There are already several embedded processors out there that directly execute Java bytecode, as well as Sun’s own picoJava specification for same. Performance for these is said to be less than that of a standard JVM running on -say- a high end x86 or PPC. Perhaps Sun could let the other guys chase the embedded space and build a processor that goes directly after the desktop & server space.

   Betcha that StarOffice would run pretty well on it.

8. fun fact: sun’s largest contract is with AOL. 60% of all sun servers go directly from the assembly line into an AOL server farm.

9. I’m not sure if Sun selling a 33 year old OS is quite the point.

   Perhaps it’s all the *free* OSs that are based on the (damned solid) 33 year old OS that are taking their toll; add to that the inexpensive PCs that they run on, which replace Sun’s proprietary boxes, and the picture comes into focus.

10. So what would I do with Sun?

    Note that there is no synergy between their three elements of OS, Microprocessor and Programming Language. No synergy means that Sun has to compete on OS with OS suppliers (Linux, Microsoft, …), on Microprocessors with Intel, AMD, … and Language with IBM, BEA, Microsoft, Open Source, …

    The answer is 1) invovate and 2) create synergy (to narrow the competitive field).

    So build a Microprocessor that is specifically designed for it’s primary use – to be programmed in a high level language. The OS needs to be an integral part of this Processor/Language combo. And finally, how about a language that uses some of the research of the last 20 years, instead of thirty years ago – a language focussed on Programmer Productivity.

    Sun needs to be bold, or forever pushed into a niche market.

11. Alex: you’re scaring me now. The product that you’ve described is the MIT Lisp Machine, circa 1980. There was a custom CPU, optimized for running Lisp (tagged data, bounds checking, etc.). The operating system was written in Lisp. And the language and environment were all about programmer productivity!

    Sadly the shareholders of Symbolics, a company created to commercialize what had been a federally funded research project, did not fare very well…

12. Sun has a dominance on the server side, but is poor on the client side. All applications are limited to HTML and applets are not very successful. Macromedia with its light weight client is set to give better experience on the client side. Sun should team with macromedia to integrate its desktop environment with flash/central and go ahead with its $100/desktop plan.

13. Philip: I mentioned that they still give away Java for free, so obviously they won’t be making money selling just the language. They never have.

    What I think Sun is trying to leverage is the acceptance and popularity of Java in the business community to reduce the commoditization of their products. If companies could mix and match solutions with different java application servers, OS’s and hardward, and never have any issues, then they’ll buy the cheapest option for each component. But the reality is that a lot of time and money is wasted trying to get all this configured and working together smoothly. If you can buy Sun’s solution and avoid all that trouble, then maybe you’ll buy a Sun server with Solaris where otherwise you would buy something cheaper. I think that’s the thrust of the their strategy.

    I’m not convinced it will save Sun, but I think it’s smarter than trying to sell to individual consumers, which is not a customer than Sun understands very well since they have always focused on the high-end business market.

14. Ben said: “that business alone does not provide a basis for Sun’s current valuation”.

    What does valuation have to do with anything? The market sets the value, all too often independent from what the company is actualy worth. While the company should grow after an emision to be able to pay dividend and eventualy buy back stock, if someone buys a share not during an emission both from some other trader and pay too much, that is their problem.

    Stocks are evil for that reason. Bonds (effectively a loan) are much better. The company has to perform well to pay them back. Where as with stocks, once they get the free money, they can do whatever they want. And because the (hired) people running the company own shares (or options), they have a conflict of interest. All too often companies are run on “share holder value”, and run into the ground. The new hotshot CEO lays off many people, including sales and R&D. Net result? Next year there is massive profit and the CEO cashes in and moves on a hero. Two years later it turns out that the few sales people left have only been able to find time to sell to existing clients and not new ones and that the competition’s healthy R&D department has the next best thing while their own has barely been able to keep existing products up to date. Oops.

15. Philip, as I can recollect, Sun tried to do several years ago just what you have proposed but on a smaller scale – I mean office networks, schools and libraries networks etc. – i.e. one central Sun server and a lot of relatively cheap terminals for users. However, it appeared that all these facilities can be built and supported cheaper and easier on the basis of casual PC boxes + Windows or GNU/Linux software while giving more fun and potential opportunities to the end users to boot. Your proposal also reminds me partly of Iridium. This project promised reliable communication via satellites between two arbitrary points on the globe and targeted those who might need such service. No much difference from accessing your data from wherever you happened to be. Had Sun ever embarked on your proposal it would have ended up exactly like Iridium but most likely without military barging in an buing up the remains of the project. As to the completely new infrastructure for computing, I think, there might be better approach targeting much larger share of potential users. Let’s see what people already have had and what they might want in addition.

    The facts on hand are:

    1. nearly any person dealing with computers in any way has already got some data and, consequently, some place for its storage. What is necessary is just the free unbarred access to this data from wherever one happened to be and by means of any standard hardware and software one chanced to possess;

    2. equally, in most cases one must have had some hardware and software he calls his own including communication harware;

    3. most computer activities of ordinary people are not mission-critical;

    4. broadband access to data is desirable in principle but even much more humble opportunities will answer to the most part of the really important problems one routinely faces day after day.(By the way, in Japan local folks are loathy to get on the new broadband standard for cellular phones. They are quite content with what they already have.)

    5. people want to be able to access and share
    freely data of any kind and surely need some degree of privacy and confidentiality doing this. It looks like they will crave for higher degree of privacy pretty soon en masse out of obvious reasons (RIAA’s recent moves build up an astonishingly wonderful preconditions for this desire).
    From the aforesaid it follows that the new computing infrastructure in question people may want to have around is nothing less but a complete network which is to be partly alternative to the existing Internet (i.e. self-sufficient to some extent and free from monitoring and anybody’s domination) and partly integrated with it wherever and whenever it is possible and useful
    without compromising the basic ideas behind this smaller network. The final structure would be highly heterogenic in nature.
    Is it really possible to build up such a network? I think so. Two basic preconditions for this have already arrived:

    1. now this time more and more household and office networks go wireless;
    therefore, what is left to do is to write a software allowing your computer to speak privately to your neighbour’s computer next door or across the street or from the street outside where you happened to be to a computer inside of a building or a house in front of you.
    (There is already in the USA the practice of warchalking – free WI-FI Internet access from some fixed points in the streets wherever people living over there want to give such an access to others, as far as I know.)
    In addition, the central wireless hub station
    which already makes your home or office wireless is to be made able to scan around in search for its counterpartners in the same manner as a cell phone is searching for available retranslators. Upon finding them, their addresses (see below about the address system) are to be retrieved and the chart of all availabe connections is to be built-up.
    This chart can be updated automatically from time to time.
    2. there is a global GPS system which allows to measure the coordinates of your computer with a precision up to several meters. This would be your address on this net. You don’t need any committee anymore to distribute domain names and superwise this procedure. You don’t need any provider anymore to give you your address and to superwise your activities after. Your address is mobile in addition. Thus the whole network would be a mobile network by default. And the address space is virtually unlimited. More independent GPS systems are to arrive soon. Europe is developing its own. This will make the alternative network even more independent and reliable. In addition to the GPS coordinates each computer may bear also its arbitrary unique name at owner’s choice. This will allow it to sign up to the net if necessary and make additional opportunities, similar to the roaming in the case of cell phones for example, available.
    Now, let’s imagine you are somewhere in the street and want to retrive data from your home computer, for example. You set up any box you have on hands and send a request to your home computer. This is an easy example because presumably the GPS coordinates of your home are known to you. This request contains just these coordinates and your current coordinates. The nearest available computer of the net looks up its current connection chart and figures out the best way to route your requests to the right geographical direction through this alternative net if possible or through the main Internet if it’s faster or more convenient our of some reasons. Actually, the software in your box can negotiate all these issues with the network devices as to by which way your request is to be processed depending on the desired degree of confidentiality, speed, bandwidth etc. Upon receipt of your message your home computer will reply in the same manner.
    There are many details to be discussed yet but, in it’s core this proposal seems to be feasible and socially useful. What is important, this net, even being introduced on a small narrow-band scale, will make the main Internet much more free and secure. Nobody would probably plank down money and resources on monitoring users Internet activities if one knew that the most critical and private data are likely to be delivered through the alternative net which can hardly be monitored.
    As to the Sun, it can surely find its place designing and manufacturing hub stations (mini-servers) for this net. Probably, these devices could be native Java computers or have this Java embedded in some way if Java is so dear to the Sun’s heart. Sun can also continue to write some exotic software for this net and this would be really exotic software because this project offers a lot of room for creativity. In general, this proposal goes in line with what Sun is already about.

    (P.S. This net will open new ways for file-sharing obviously. Thus, search and request of files can be performed privately through the alternative net. After this the requested encrypted files are sent to some E-mail address through the broadband main Internet.)

16. Sun should merge with Apple and port OS/X to Sun’s server platform, then offer it as a free upgrade to customers to ensure adoption. Make Java part of OS/X and have the server edition include a superb J2EE application server. Then merge with Google and integrate their search capability with the desktop platform, the server platform, iTunes, etc. Then buy AOL from Time Warner, and go toe-to-toe with Microsoft…

17. It would also help if this new company would buy Borland or some other company that makes a good Java IDE.

18. Andrey, Sun is still trying that. (servers with thin clients) And if companies would adopt it, it would be great. User’s workstations would be reliable, all the same and you’d save a lot of money in support. The first problem is that this only works if an enterprise does a complete and total move to that platform. That is an almost impossible mission. For start-up companies, the inital costs are prohibitively high to implement it on a small scale.

    That and the psychological lock-in companies have with M$. What I mean is, they think they cannot operate with the outside world without Office and Outlook, not realising that 99% of their document exchange is internal (and that an editable format like Word is a horrible way to “publish” your documents) and that Outlook is just another email client.

19. Philip: Think also of the Dorado. Dr. Alan Kay said “…current microprocessors are 32,000 times faster yet only 50 times more powerful than the Dorado… The microprocessor industry is focussed only on sub-goals, like line-width, instead of the real goal…”

    Analogies are risky. They often backfire, but here goes:

    Think of (clothing fastening) Buttons and Velcro. When Velcro was invented, it was all the rage. It was used in all sorts of (inappropriate) areas. Eventually we settled back down to the better, older idea of Buttons with a few judicious applications of Velcro.

    Perhaps we have that situation here with Microprocessors, OS and Programming Languages? Hopefully we will settle back down to the better, older ideas?

20. Bas: Like it or not, corporations have a duty to try to maximize value for their shareholders. That doesn’t mean they have a sense of obligation to bail out investors who bought a stock at the peak of an irrational bubble, but it does mean that their objective is to try to grow the business and make the company more valuable. That means trying to increase sales while also trying to reduce costs, often to varying degrees depending on what they are able to accomplish and how much competition they are facing.

    The value of a company is the sum of its assets and the present value of its expected future cash flows. If a company shifts strategy and basically says, “we won’t be growing much anymore” their stock price will fall to adjust to the new expectation of the future cash flows. This might happen eventually anyway, even if the company is trying to grow and they fail. But that is a more gradual process as the results fail to meet expectations quarter after quarter, and the expectations are adjusted slightly each time. So they have to face reality each quarter, but they also have to keep trying to meet expectations, and they have to follow a strategy that has a good chance of getting them there.

    Sometimes companies finance the things they are trying to do by selling stock, and sometimes they take loans, either directly from banks, or by selling bonds. The combination of these is a company’s capital structure, and during different market conditions it is advantageous for company’s to have a different balance between equity and debt financing. If the balance was shifted entirely to debt, as you suggest, there would be a lot more bankruptcy and very little innovation. Debt is risk-averse, so startups would not qualify for loans, and they would never get funded. Existing companies that had a bad quarter would have to declare bankruptcy to pay off a loan, so they would not do anything that doesn’t guarantee them a consistent income. So they play everything safe and borrow as little as possible. Sun in that situation would fire most if it’s employees and just sell existing designs until they could pay off their loans.

    It takes equity financing (a stock market) for innovators to get funding for risky ventures, or for existing companies to have the freedom to take risks which justify the existence of many of their employees, such as their entire R&D staff.

21. I just saw the price Microsoft charges the Bank I consult for SQL Server 2003 – $463 CDN
    per year per server!! How can Oracle or IBM compete?

    Perhaps Sun should move its head quarters to India and focus on the outsourcing business – like IBM.

    In order to compete with Microsoft all of Sun’s parters (Oracle, SAP, & Siebel) are being forced to squeeze everything except for their profit sanctuaries.

    Oracle with its Oracle Parallel Server on Linux
    SAP with Linux and it’s MySQL investment

    IBM has the only model that works in competing with Microsoft. Become the customer’s interface with technology (outsource partner) and take a chunk of change no matter what they implement.

    Anthony
    More on software & hardware commoditization:
    http://xminc.com/mt/archives/000019.html#000019

22. Channel Stephenson:

    1) [Deleted]

    2) I’ve always wanted a tattoo watch. Invent a tattoo dye made of LCDs. Pour money into subcutaneous computing — the next next embedded/ubiquitous computing. Powered by induction batteries, or possibly even the natural cellular sodium processes. Tattoos become watches, PDAs, geolocation services, advertising, and/or makeup. (Note to Legal: Buy off Neal Stephenson’s patents.)

    3) Again, steal from Stephenson, and start up ractive entertainment selling both the servers and machines to create interactive movies, the servers to run them, and the servers to morph/translate ractors performances in real time. Start off in Vegas, Tokyo, Hong Kong, …

23. I see a lot of technical solutions here but no real “requirements” or “needs” Sun customers could express.

    A person or company don’t “need” a technology for itself, but they could need what it provides. This sentence seems obvious, but when talking about solutions to Sun problems, I only read “Sun should port OS/X to Sun’s platform” or “Sun should team with macromedia”, without analyzing what could need a customer that only Sun can satisfy with great success.
    Does an individual need WAP ? No. He needs useful services. Does an individual needs Software, Hardware and Networks ? No. He needs means to get and publish useful content. The term “useful” here is more important that all other terms because “useful” is to be taken directly or indirectly as “useful to achieve my Maslow needs”.

    Remember Maslow Hierarchy of Needs :
    1. Physiological
    2. Safety
    3. Love
    4. Esteem
    5. Self-actualization
    The order is important, you can not satisfy need N if need N-1 is not yet satisfied.

    Which Maslow needs could be directly or indirectly satisfied by a global integrated electronic infrastructure and services provided by a company like Sun ?
    Sun, Apple and other “creators” (as opposed to “innovators”) currently makes money more from the “esteem” need than any other Maslow needs.
    Some could say people buy Hardware and Software to satisfy the “safety” need. Who has never heard something like this :”Buy secure and innovative Microsoft or Sun products instead of trying those risky open source software created by unknown people!”. The fact that buying a Sun product is considered a *reasonable* choice by Sun customer is not here to satisfy the “safety” need but to satisfy the “esteem” need, because they want to show everybody that they are “reasonable” and make “reasonable” choices and that we must all esteem them as *reasonable* (or *good*) persons.
    If people begin to see technologies like Java and Solaris like good old 30 years old technologies, they will reappraise them. Maybe we are at a time where people are considering using technologies to satisfy other needs like “love”, “safety” or “physiological” (“physiological” needs can be satisfied by technologies if using a technology gives more time and money) or even “self-actualization”…

    Maybe all theses thoughts are nonsense or just a try to satisfy my “self-esteem” need.

24. Ben, I agree with a lot of what you say. Equities allow for innovation, true, but outsider investment does not mean an IPO and being publicly traded. Wealthy individuals and companies in it for the long term are a much better solution. Even when publicly traded, a company with a healthy bank balance (which sun certainly has) shouldn’t care too much about the share price, they have enough money in the bank to innovate and create _long term_ value. Luckily, according to Scott McNealy in a recent interview, that is his main concern, not daytraders. Come to think of it, what do many succesfull tech companies (MSFT, Oracle, Sun, Google, Yahoo, Amazon) have in common? They all have their founders at the helm, not some hired MBA yealding hotshot navigating the company from quarter to quarter, maximising the value of his portfolio. (I believe the owner of this blog can testify to what a bad idea that is)

    “So they have to face reality each quarter.” Another good way to ruin a company is to run it quarter to quarter and when one doesn’t go great, change direction the next. Mind you, these quarterly obsession is a purely American thing, European companies and investors seem to take a much more realistic approach. That doesn’t mean there have been some dissasterous dot-com IPOs here in the late 90s, though. Probably worse because there is no such thing as a quiet period and companies can actively advertise to the public they should buy their stock.

25. Bas: I don’t disagree with any of that. The American focus on short-term results is a serious problem, especially as it affects decision-making.

    The main argument I was trying to make was not that Sun should be focused on the short-term share price, but that they can’t give up strategies that could lead to growth in the long run. They could be much more profitable tomorrow by scaling back to just selling hardware, but that business alone will be shrinking over time, so it would be a bad move. They need to sell a bigger package and address more of the fundamental needs that Jerome wrote about, and address them with end-to-end solutions that give them higher profit margins than the separate technology elements of that solution could give them separately.

    Sun’s biggest asset right now is Java, not because it’s any better than Phil’s favorite languages of the 70’s and 80’s, but because businesses are comfortable with it and are willing to buy solutions built on it. The more Sun can work that to their advantage, rather than letting their partners get most of the benefit of the goodwill towards Java, the more viable they will be.

    Phil’s AOLserver solution was and is great, and it really bothered me that it wasn’t more widely adopted. It was kind of sad the way the VC took control of ArsDigita and shifted to developing a J2EE product that still has not gained much acceptance. So I understand not everyone loves Java, and I personally see it as just another language with pros and cons, but I happen to like it and think it’s pros form sort of a sweet spot for a lot of problems.

    Phil wrote in his PandA book about the problems with developing in C or C++, and that mainly IT customers don’t want to take on that responsibility themselves, so they buy software from Oracle and Microsoft, where thousands of programmers work for years to develop stable products written in C. So Phil argued that big companies put all of their important data in solid ACID-test compliant databases like Oracle, and write their business rules in SQL or PL/SQL.

    I agree with that, but I would add that today they are also willing to build applications in Java, and that is the best asset that Sun has.

26. Interetingly, we aren’t the only ones having a discussion about Sun’s future strategies: http://www.internetnews.com/bus-news/article.php/3086651

    I also have to eat my words about Sun’s thin client solution being to expensive for small businesses. I had a look and less than $25000 buys you a pretty big server, 15 clients (the 15″ LCD ones) and all licenses. That is about $1500 per workstation. And remember, the office suites are free and you can run many free mail solutions as well. Compare that with going for PCs and you are close to $2000 per workstation and window/office licenses. And then you have to add a server. And PCs are a pain in the ass to support. Would seem that even without MS licenses, running Linux workstations would be more expensive! (and much more of a pain)

    Still beats me how to capitalize on Java, though. Java is just a different kind of language than Tcl, both very good, but for different purposes. (although than play quite nicely together as well!) Phil has some healthy ideas about development (it was photo.net/wtr that gave me a great start when I decided this thing might be fun), his passionate rants against Java are a bit over the top, though!

    The thing about getting companies to use somebody else’s database and writing the application is “simple” languages is that not all applications would work well inside a browser and not all scripting languages are good at GUIs. Tk is getting quite good these days, but a great big desktop app written in it, especialy on a multi-developer team doesn’t look like a lot of fun to me!

27. Compare that with going for PCs and you are close to $2000 per workstation and window/office licenses

    Where do you shop? Try 2.2 GHZ, 128 MB RAM, 40 GB Hard Drive 15″ Flat Panel, Windows XP, MS Office, 48x CD ROM Drive, Integrated Ethernet from Dell at $899 (save $220 for CRT Monitor)

    And we run our business 24×7, zero down time for 5 years on MS Windows Servers. So try a 2 GHz, 512 MD RAM, Windows 2003 Server (from Dell) for $998.

28. About Java technologies : how about beating Microsoft at its own game ? Sun could *sell* a 100% compatible Java compiler to generate .NET bytecode instead of JVM bytecode. As it is oriented toward an expensive platform as Microsoft Windows, Sun could sell (instead of giving away) the Java compiler for .NET .

    Unfortunatly, great people have better ideas. Look at this one :
    The SnippetStore Project
    They emulate the different J2EE APIs by reusing .NET Framework libraries ! And they manage to run the complete Java Pet Store on .NET. Sorry Sun, you can’t sell Java.

    About the professional market : Maybe it’s time now to make large-scale savings by mutualization of information systems (IS) infrastructures. When it comes to one information, all software functions you need are:

    – create the information

    – modify the information

    – delete the information

    – search the information

    (and a batch mode for each one). When your information is stored into database tables, all these functions can be automatically generated ! (See Oracle Forms, or Oracle WebDB). You don’t need Java for manipulating tables in a database. Add to your information database a Business Process Engine (just a tool that interprets a screen transition program described as an information into the same database). With a BPE like that your program becomes data instead of source files you must maintain (and must know the language used in the source files). The BPE shows you one of the 4 functions above when it needs a user interaction and then execute the program : a series of conditional calls of functions taken from the 4 functions above where data flows from function to function. If you need special functions, e.g. creating a PDF file for invoicing, calculating a special mathematical function on a particular information, sending an information via email or even snail mail, all these special functions can be seen as black boxes with inputs and outputs. These black boxes must be developped once for all and always, and be reused as services offered by the IS service provider. In short, for its Information System, a company don’t need to develop software anymore except software tailored to special domains needs. It should only focus on its data model and its business processes. The IS service provider could be Sun. But it’s probably Oracle or WebMethods that could easily make the move because Java like any other language will not be useful anymore as business processes can be expressed via user-friendly graphical schema. This bring another problem : what to do with developers ?

    (to be continued…)

    PS: Don’t tell IT managers the 4 functions can be expressed in a good old 1970 technology : INSERT, UPDATE, DELETE, SELECT.

29. “Come to think of it, what do many succesfull tech companies (MSFT, Oracle, Sun, Google, Yahoo, Amazon) have in common? They all have their founders at the helm, not some hired MBA yealding hotshot navigating the company from quarter to quarter, maximising the value of his portfolio. (I believe the owner of this blog can testify to what a bad idea that is)”

    What MSFT, Oracle, Google, Yahoo and Amazon have in common is that they have products/services that large markets want to buy. Sun does not seem to be able to make that claim anymore.

    Java is so fragmented and has so many different implementations that its cross-platform abilities are rendered irrelevant. Besides, it is notoriously difficult to make money out of programming tools – particularly when you are giving them away for free and they don’t require the purchase of your software to run them like Java (use Microsoft.NET as a comparison).

    I can’t speak much for their servers, but it doesn’t seem likely that Sun will be able to compete with PCs running Windows 2003 Server or Linux, except in a few big companies that are diehard Sun customers.
    Ie this news just came in to me: “Friday morning, Hewlett Packard (NYSE: HPQ) announced that it would pay Sun customers $25,000 to switch to HP computers running Linux OS freeware.”

    I agree with Philip – it is not clear how Sun’s current market valuation will be maintained without products to sell and interested markets to buy those products.

30. We’re ripping out Sun boxes and replacing them with Pentium-4’s running Linux. New Sun boxes and Solaris are prohibitively expensive. The “lintels” run us about $800 each. Comparatively, benchmarks rate the P4s as about 25 times faster than our old Sun boxes. We could never afford Sun boxes as fast as the P4’s. We’re happy as hell in Lintel land.

31. what is the business need Sun fills with new boxes? what is the new stuff that Sun releases the everyone wants, but no one knows they want? Philips is making the number 2 argument, others (jerome etc) are talking about the first one. Sun probably needs to do both.

    As to business needs, Sun needs to implements SOLUTIONS that allow corporations to reduce their staffing levels while providing the same or greater service levels. This is what java is supposed to do. More and better programs written by fewer programmers. (personally I find python is way more efficient for my productivity than java ever was). Sun needs to do more of this. No corporation is going to choose an option that requires hiring people, but many will choose options that allow them to downsize.

    as to the “new stuff” approach I think Philip is close here. Sun is “big iron” No one I know buys Solaris unless they are a corporation. I know of no small businesses that do today…but I know of a few that have gotten rid of Solaris because Linux is equivalent and cheaper…in fact, Linux is probably better because it has a faster development cycle.

    so my solution would be to ‘zig where they zag’ (dw trdmrk) Sun needs to target the “young entrepreneur/college student” market. Provide tools and technologies to this group of people to let them develop businesses to eat the lunch of the older more established companies.

    for example: If Sun has lots of installs in banks/cc companies (which they do) then Sun should know lots about this industry. Wh y not setup some Sun servers that handle all the regulatory requirments for doing banking transactions, provide all the tools to interface with “SunBank” and then release software at colleges to let students setup their own “microbanks” ala mini paypals. Sun provides all the application interfaces, and stores the money, but the minipaypals handle the transactions…even micro transactions. Sun charges penny fees for transactions. A student charges penny fees for transactions. And we’re still in the realm of real micropayments…at micro bank levels…in universities where micropayments would probably be very successful.

    you know if you could order a pizza through your browser and pay for it through your dorm rooms bank acct that also donates a penny to “insert college charity” when the transaction happens college students would do it.

    anyway, that’s just an idea…distributed banking, running on Sun hardware with Sun api’s. the point being you have to get “cool new” abilities to people “without” abilities but “with” drive and ambition. look what college students/peer to peer did to the music industry? what do you think peer to peer banking would do to the banking industry? People who don’t know what they are doing come up with great solutions to problems…isn’t that how Sun got it’s start anyway?

    -calvin

32. Hey Philip, great blog. I’m one of your frequent readers. Well, would you be so kind to change your RSS feed in that the links would direct to your post instead of a link mentioned in your post? I just clicked the link for this post in my RSS aggregator (RSS Reader for Firebird, btw) and I got directed to the NYT website hehe… Well, just wondering… Keep up the good work 🙂

33. AC: Just that Linux servers serve you fine, doesn’t mean they do for everyone. Some apps simply do run best on massive 64-way multiprocessor machines that support terrabytes of diskspace and 64GB of RAM. And once you are in that business, you may as well go all they way and have all services running on one platform.

    While you can say that distributed computing is more effective, you are probably right. But you can buy a machine described above, install Oracle on it an run happily, reliably and supported. For distributed computers, you are on your own. That is nice for scientific projects, but not for most businesses.

    Not that Sun’s machines are mainframes, but I remember a quote from someone countering an argument that mainframes are obsolete: “Mainframes are being used by thousands of obsolete companies, serving millions of obsolete customers and making obsolete billions for their obsolete share holders” (or something to that effect)

34. Bas: I like that last expression. The shareholders are indeed obsolete in the United States because now we’ve figured out ways to pay all of the profit to the managers 🙂

35. AC: The founders at the helm comment was regarding equity investment and the problems that come with it, not a way of saying the Sun will be alright because their founder still heads it.

    AP: You would make people work with Office/Windows XP on a 128Mb machine!? You massochist! 😉 OK, so machines can be had cheaper. But buying costs is only part of the cost to operate them. For a small set-up example I gave, that probably wouldn’t make much difference as you still need 2 support people to cover each other. But if you have a 1000 workstations, paying $500 more per WS, so $500,000 extra investment. But instead of 20 IT support people, you only need 10; you’ll make your money back in the first year. And while the hardware depreciates in the next two years before replacement, you save $1M! Makes perfect sense to me…

    And even for smaller companies, you do away with the headaches of virusses and people running everything they want left right and center and wasting productivity.

36. What to do if you’re Sun? Quit.

    Liquidate the company, sell the assets and turn over the proceeds to the shareholders (minus generous payouts to the employees — along with rights to any patents they had anything to do with).

    Why continue a losing fight until the $6B in the bank is gone out the window? If you have nothing better to do than go home, then you should just go home, damnit!

    Will Sun fare any better than Netscape (whose best product has been released open source and free to consumers)? Java = free application platform that will not produce revenues or profits to the levels needed by SUNW to survive. Better to call it a day with your head held high rather than continue to fight until you are overtaken by rivals (for sports reference, see Holyfield in boxing).

37. Cringley had some insight in Feb 2003 on this.

    How to Avoid the Almost Certain End of Sun Microsystems :
    http://www.pbs.org/cringely/pulpit/pulpit20030213.html

38. Vintage Clothes

Comments are closed.