Enterprise Outlook "A Business Conference from Technologic Partners" The Future of Intranet Computing June 30 July 1 1998 San Francisco airport Marriot Hotel URL for Organizers: http://www.tpsite.com Trip Report by Geoffrey Fox This conference is one of a set held approximately every 3 months by an excellent organization who publish the very informative newsletters ComputerLetter (approximately weekly) and VentureFinance (monthly). ComputerLetter is very technically savvy and I can strongly recommend it as a source for the latest in the computer industry -- especially in startup arena. The next conference in this sequence will cover Internet Commerce in September. Although I was essentially the only University (research) representative at this conference, I found it extremely valuable and I believe many others of our community would have also found it beneficial. The conference featured an intense schedule with no breaks in a 8am to 7pm schedule except the 90 minute lunch session which include a significant talk. The last 2 hours of the formal program were however a demo session with appropriate refreshments. Interestingly whereas most of these activities dazzled one with the professionalism of the participants, I thought the demos not as carefully prepared as typical in "our" community. There were 500 attendees from the software industry, press and venture capital communities. The conference was exceptionally well organized and not one session ran over by even a few minutes. An interesting facility was the "RSVP voting system" which allowed audience to answer multiple choice questions with instant feedback on voting results. Although I was an outsider, I did meet colleagues I knew from earlier days in nCUBE and Thinking Machines. Gary Smaby -- a familiar figure in HPCC, especially SCXX, was there with an interesting startup CARTIA. The conference was valuable as it reviewed nearly 100 small software companies in the Intranet area and so gave one a very good view as to where industry enterprise software systems could be in the next couple of years. The conference included plenary sessions with "Industry Leaders" (such as Andreessen from Netscape, Kahn who founded Borland and Siebel from rapidly rising Siebel systems) and panel discussions on such topics as "which are most promising applications", "best ways to integrate applications", "futures in marketing automation", and "picking winners from showcased companies". The 100 small businesses typically had 30-100 employees, around $10M venture capital investment and had been set up in last 3 years. These companies were mainly in revenue stage (selling product but unprofitable) with 10% still in product development and 10% already profitable. They were presented in terms of 20-minute talks and for a third of them, participation in the evening demonstration sessions. The rest of the report is a mix of sound bites and comments on general issues where we will start with Java! Java is a Central Technology Java is viewed as an uncontroversial and important language -- it is not viewed as a fad or likely to be replaced soon -- a comment I hear from other circles. One interesting RSVP survey on most important influence on industry in next two years rated Java in second place (19%) trailing Microsoft (57%) and equal to application servers (19%) but way ahead of Netscape and IBM at 3% each. Most companies (Marimba was an exception) built around a "thin client" with development on the server -- not client. The same trend was equally obvious at JavaOne conference last march in San Francisco. This is in some way inconsistent as most of the compute power is in clients and not servers and server- based solutions have scaling problems. However for corporate customers, the focus on servers can be understood for 2 reasons. 1) Corporations find it very hard to manage clients and so "pure server" solutions is most attractive to an overworked Chief Information Officer. 2) More cynically, one can quote Andreessen who says, "the situation with client side Java is grim" with "no standards and poor performance". He recommended JavaScript and dynamic HTML on the client. Some of us think that this follows from Microsoft undercutting business model for browser market with Internet Explorer being free and so we now live in a world of inexpensive low quality browsers. 3) Note there is another reason (the "webTV" reason) for consumer market to like thin clients. These are what the broad public can afford. However this is NOT why corporations use thin client ("4-tier") architecture. Indeed Andreessen predicted that home PC prices would continue to drop and become like mobile phones -- given away free with some sort of value-added service. I asked several companies who appeared to have "thick clients" (a.k.a. significant Java applets) how they addressed the "grim Java client issue". Most responded that they were (or will) using Java applications and not Java applets. Data Visualization Exceptions to the above analysis were some companies successfully using Java applets to support visualization frontends for datamining. AlphaBlox and Brio demonstrated approaches to "drill down" thorough data with additional refinements added dynamically by user. I think some of these user interface ideas demonstrated for data stored in conventional relational databases, could be applied in scientific data analysis. AlphaBlox showed a neat way to build up custom visualization interfaces by dragging and dropping Java components into an HTML editor like FrontPage. Marimba -- Client Side Management An interesting company is Marimba whose CEO Ms. Kim Polese gave an excellent and charismatic presentation of their Castanet system. This was viewed as a very interesting company and their 20-minute presentation was very well attended. However there was only modest enthusiasm in the final session for the likelihood of Marimba's success. This dichotomy can be explained by the emphasis on thin clients and the drastic cooling of enthusiasm for the much-hyped "push" technology. Marimba is now emphasizing ADM (Application Distribution and Management) as done by Tivoli in the traditional business world. Their technology allows one to deploy with version control dynamic data and programs on clients. As I think "thick clients" are obviously going to eventually win (default PC's will soon have gigaop processors and gigabyte memory), ADM seems to be to important in the long run but we will see if there is enough near term interest to allow Marimba to flourish. Java on the Server There were a set of companies building server side "glue"/infrastructure. Three presenters (WebLogic, Persistence Software and Novera Software) were offering various flavors of Java Servers. I pressed them to distinguish themselves and in a nutshell one can highlight WebLogic for its JDBC driver roots and very fast server, Persistence for its interesting object technology and Novera for management capability. Persistence builds the by now "party-line" Enterprise Javabean data object to relational persistent store. They also had a seemingly (to me) important technology to cache at the object layer -- i.e. within the dynamic software layer. In some cases this will lead to very important performance boosts even in the simple case of web pages with the growing tendency to use sophisticated database backends. Finally Novera offered expertise with a set of capabilities to assign services to servers. This can support with appropriate security, replication needed to use Java servers to run very large distributed enterprises such as a retail store chain. However they did not support the type of dataflow capability built into NPAC's WebFlow interface to Java(bean) servers. Application Integration As well as these tool companies, there was in fact more interest by the venture capitalists in a set of "application integration" offerings. These technologies are motivated by the use of some 35 independent significant applications in different parts of the operation of a major corporation. These need to integrated together to create the "business ecosystem" for the most competitive information rich companies. In this sector, companies included SmartDB which produced near real time transformation tools to go from one application to another. TIBCO, CrossWorlds Software and Software Technologies Corp. provide dynamic middleware integration using XML and Java as appropriate. This overall market is estimated at $2B to $5B per year depending on whether one counts just software licenses or software and services. Knowledge Management There was a class of interesting companies in the Knowledge management and presentation areas where I found blurring with the concept of groupware as asynchronous collaboration tools are "just" tools to help convert data into wisdom. For example, Intraspect Software supported a mix of Lotus Notes like threaded discussion lists with tools to organize information. I found several companies with interesting capabilities but none with the complete set -- it might be useful to view these offerings as unbundled tools with another activity (company) integrating them together. For the academic community, this area is of particular interest as it resonates with NSF's interest in KDI (Knowledge and Distributed Intelligence). Verano stressed that in a typical organization 80% of their data is unstructured (outside formal databases) and 70% was on client computers. Verano and Intraspect stressed the importance of employee email and the need for its automatic organization. Verano highlighted security issues. Another piece of groupware came from Amplitude Software and supported dynamic scheduling of resources such as offices for mobile professionals, audio-video and general conference rooms. Hyperparallel (which stressed support of physical location of information in maps) was one of several datamining companies but there seemed no consensus as to key analytic capabilities. Relevance Technologies and Inxight Software typified the digesting of unstructured text but perhaps the most innovative approach was a set of companies (CARTIA and Perspecta) that mapped information into 2D or 3D "knowledge spaces" where hills represented accumulation of wisdom. This approach to unstructured data contrasted with the drill down methods and relatively conventional if dynamic charts of AlphaBlox and Brio using Java applets to display information from databases. This whole area did not seem to catch the attention of venture capitalists. Eventually a suite of such tools will be on everybody's desktop and used to train their mobile agents. Training There were a few distance education systems presented with Centra and Placeware both painting an optimistic picture with industry currently expending some $60B on training. This does not include either DoD of the education (K-graduate) community. Docent was an entry in the asynchronous tool area but their demonstration at least did not show great sophistication. Centra and Placeware however have sophisticated core technologies, which are related to those used in NPAC's TangoInteractive system. Note that Placeware recommends using analog phone lines for audio as they claim that Internet does not give sufficient quality of service for digital audio-video conferencing. Sound Bites (Random Snippets) 1) Kahn's new company Starfire is building wearable computers. Soon you will be able to install a web server in the heel of your shoe and use it to control your heading aid. The company has an average employee age of 26 and both chief technical officer and VP for engineering are under 30. One needs to keep forming new companies to preserve these parameters. In fact established companies suffer from the serious disadvantage of recognizing weekends. 2) In 1995, the largest independent software vendor was Cullinet with $200M in annual sales. Just 13 years later, Microsoft is 50 times larger in sales (and even more in market capitalization). There are many companies with over $1B in sales. This represents a major largely unheralded change in industry make-up. 3) We learnt that BMC ($11B market valuation) has a revenue of over $300K per employee and pays from this an average annual salary of $85K and symmetrically makes a profit of $85K per employee. Such is the good life … 4) We learnt about Siebel Systems with a market valuation over $2B (15 times sales) which is still rated a "buy". Siebel himself in a plenary talk illustrated the value of integrated information by describing the current clumsy home loan application and evaluation mechanism. This was compared with a futuristic view of how this would be done from one's laptop after breakfast. Siebel imagined a world of personalized software agents specializing in buying and selling while briefing agents will abstract information for you. 5) Filipowski (PLATINUM technology) added to Siebel's vision and hypothesized that in a few years, the internet would as essential as electricity is today. 6) There was discussion of the "Death Stars" -- Microsoft Oracle SAP - where you were doomed if you directly invaded their territory. However there was plenty of opportunity outside their tractor beam. In this regard it was noted that the "fear of Microsoft" was a more serious problem than Microsoft itself. Microsoft was claimed to be "NT Centric" and therefore not able to focus on broader issues. 7) I noticed that there appeared to be a minimum starting price of $50K for products. It appears that industry accepts this. 8) The presentations by show cased companies were judged to be organized by either "technology out" or "marketing in" principles. The latter was recommended if you want to raise money but some frustration was expressed for the stealth technology mode where no information was given on technology approach. Note that today technology is advancing much faster than the understanding of the buyer. Thus you cannot sell on the basis of technology but must stress return on investment to the customer. 9) There was an interesting discussion of marketing to the non-corporate consumer and the difficulties of doing things either directly or through Internet service providers. 10) One needs to integrate "Inbound Logistics, Operations, Outbound Logistics, Marketing and Sales, Customer Service and Support" in one's business ecosystem on to network of "Extranet (business to business), Intranet (internal) and Internet (external customers)" 11) It was noted that as sales of Quicken software decline, the role of the http://www.quicken.com web site increases. Ariba Technologies (125 employees, equity investment $13M, valuation $113M) CrossRoute Software (50 employees, equity investment $14.5M) Epiphany (35 employees, equity investment $12M) Marimba ( 125 employees, equity investment $18.5M) Ariba was the most popular of the showcased companies for both venture capitalists (in last panel) and the audience (based on RSVP voting). Generally the electronic commerce area is very hot and Ariba specializes on the business to business side of this by supporting the procurement function. Epiphany which supplies marketing automation software and CrossRoute which supports linkage of businesses joined Ariba and Marimba as top 4 companies based on interest shown by attendees.