|HOME||PUBLIC LIBRARY||ANOTHER PERSPECTIVE||INFOPERSPECTIVES||CONTACT|
IBM wants to be your mashup company. Of course, it helps if you know what IBM means by mashup. It also helps if you prefer to pay for development and testing services sold by Amazon. Try not to let the free starter services offered by Yahoo, Google, and Microsoft's Bing distract you. IBM might be onto something good. And if you have to pay a little extra, well, what's new about that? But I'm getting ahead of the story. So, let's start at the end.
Mashup is the current term of endearment for a Web site that combines information from two or more sources. For instance, a mashup might combine map information with weather information and one called woozor does this, more or less. Woozor is still in development but even now it pulls together Google Maps (and the related Google Earth) along with various weather forecasts. Visitors can get forecasts for a week for a whole country or zoom in and get the weather for a local area. There are a number of sources of maps and there are a number of places to get a weather forecast. What woozor does is combine them, or in the current argot, mash them to yield something that woozor fans think is better than either. The result, flaws and all, was good enough to get attention. Mashups let Web surfers do more, a lot more, and have fun at it, too.
Mashups also seem to be within the reach of just about every organization. They are built in development frameworks that can be used by people who are not super Web coding hotshots, at least for the relatively simple applications that serve as starters. Moreover, things are likely to get easier, not harder. The companies that are trying to provide computer services for mashups want to make it easy to get started and, to the extent they can, keep it easy no matter what a developer wants to do. That's not to say that building mashups is as easy as, for example, writing a letter using a word processor, but some of the examples used to ease users into the systems are not much more difficult than, say, building a graph using a spreadsheet program. IBM offers an intro to its system on YouTube that makes the whole process seem easy.
While IBM's YouTube example is pretty stupid, it does include some of the elements that make mashups a hot concept. In this case some of the key data is in a spreadsheet generated far from Amazon's Web and plunked into one element of the mashup. The data may have been generated on a personal computer running any spreadsheet program or any of database programs that can spit out an Excel-compatible file. But the data could have originated on any kind of system, including propriety systems like IBM's i and z machines, as long as the source system could easily spew out the required data in a suitable interchange format.
The magic that makes it possible for mashups to be built in the cloud or on users' systems is the collection of Web development tools that have grown up in the fertile economic realm populated by Internet software inventors. These toolsets are usually developed in open source colonies, often incorporate foundations based on Java and other platform independent technologies and, generally speaking work with common server software components such as Java, Apache, WebSphere (which is related to Apache), PHP, Ruby, Perl, and other packages known the world over. But perhaps the most important aspect of the mashup toolsets is that they allow newbies to get good results using a GUI that does not require a lot of technical knowledge. Mashup projects and move beyond the reach of basic GUI actions pretty quickly, but developers who put in a bit of time with a mashup system can learn what they need to know if they make a reasonable effort.
A lot of the software used to create mashups has been ported to Microsoft's proprietary IIS, and in fact IBM's cloud offering on Amazon seems to be hooked into Open code running inside Windows. Linux may be a big part of the mashup world elsewhere, but IBM's development tools live in Microsoft country as IBM's developers' download page indicates. IBM hasn't explained why it has made Windows its favored server environment for mashups, but there may be an historical basis for its situation. The first version of IBM's mashup development system was built under Windows so users could do development on their desktop computers. When IBM's subsequent development packages outgrew desktop software and grew to require the richer environment available on a server, IBM didn't want to sacrifice its library of Windows code. Now, a major and a minor release later, IBM seems wedded to Windows.
Still, whether a mashup Web server uses Windows or Linux, a user can move data from DB2, Oracle, SQL Server, MySQL, and other database management systems into mashups with relative ease. This is clearly the case when data movement can be done in batches. Things can get a bit more sticky when a mashup needs real-time data from a system that's not part of the Web mainstream. In other words, when the data is coming from a proprietary system offered by IBM, Oracle or an ISV, a user has to make sure all the software components can cooperate. If the database doesn't provide suitable connector software, the mashup framework has to offer that facility and that's where users' hopes may get dashed. Mashup system developers are not putting in the kind of back-end support effort that is routinely made by purveyors or, for instance, accounting systems. This is an aspect of the mashup world where IBM might be able to make big gains.
IBM says its mashup technologies can pull information from WebSphere, Sharepoint, Filenet, Content Manager, and elsewhere. IBM seems to like software components from Widget Factory and it will also have hooks for Cognos.
A question raised by IBM's rich but complex mashup strategy is whether customers can get results that justify their development efforts. With IBM, what starts simple, possibly with some inexpensive blocks of time on the Amazon cloud, can quickly grow into a monster as it is brought in-house or as a user attempts to hook in-house systems to cloud services.
Would the same project be easier using clouds from Yahoo, Google, or Bing? The answer, very likely, would be yes. And the cost of finding out some of the answers that lie near the surface would be zip because, unlike IBM, Yahoo, Google, and Bing are happy to have customers sign up and use their systems for free.
In addition, because Yahoo and Google do not sell servers or software and because Microsoft seems willing to let Bing have a life of its own even if it does not boost sales of Windows servers, all three services are working on ways to provide secure private isolated clouds that are rented by a single user. And there seems to be a lot of competition of the kind that ought to help end users get more value. A recent, important example is Google's success in persuading the city of Los Angeles to use its cloud services for email and other purposes. The loser in this case was Microsoft, which is going to lose a lot of Outlook and Exchange seats. The Google victory does not revolve around mashup Web sites, but it's a safe bet that Los Angeles government offices will soon be shown how to mix Google Maps with applications that support activities ranging from traffic control to transfer payments.
Where Google is now leading other companies, possibly including IBM, may soon follow. And if Google succeeds in LA, it will have shown how the USA, Japan, and Europe might be able to offset the economically, politically and culturally painful movement of technology services from the wealthy industrial companies to the emerging Asian techno powers. Basically, Google is hoping that Los Angeles will prove that there's something cheaper than a service bureau in Bangalore: self-service in Malibu.
IBM, Hewlett-Packard, and Dell have all invested heavily in traditional services operations. But cloud computing and its most dramatic face in the form of well done mashups has the potential to not only offer new kinds of computing services but also to enable politically correct economic nationalism.
— Hesh Wiener November 2009