Peter's Blog

Two worlds light years away from each other: hard and soft SOA

Peter Csontos — Mon, 04 Sep 2006 11:13:00 GMT

In a recent article in InformationWeek, "The Dark Side Of SOA", the author talks about a survey and statements of a couple of people considered to be relevant concerning service oriented architectures. He explains how and why it is hard to implement SOA, at least now, in its early stages.
There is a sentence that I would like to underline: "SOA is about continuous change". This, of course, was said by a consultant, I would call this statement "The consultants' definition of SOA", or even shorter "Hard SOA". Naturally, this approach is all about massive custom development. No standard or productized components and processes, no controlled version upgrades. Just tons of body shopping. The dream of consultants. The nightmare of corporate IT managers.
The opposite approach: "The ERP vendors' definition of SOA" or "Soft SOA". Productized Business Process Platform with out-of-the-box and/or custom composite applications for vertical and company-specific processes. Yes, this means license and maintenance fees and upgrade headaches, but I've seen both approaches and believe me, the second one is much cheaper but still nice, bright, hype and shiny.

AMR Research Completely Misunderstanding SOA

Peter Csontos — Tue, 29 Aug 2006 06:29:00 GMT

About this article: http://www.amrresearch.com/Content/View.asp?pmillid=19699&pubid=2726&custid=381817

Just to summarize: the author says that ERP will disappear in 4-6 years because customers will build their own business process platforms with the help of system integrators instead of continuing to upgrade their SAP or Oracle application platforms.

I tell you why it's not true: Copying and paste-ing custom solutions is feasible only if you have a market segment which is not mature enough to let generalized software products arise. I've seen this from inside a company that did this in the telco billing area. At the end customers realized that the company that developed their custom solutions quickly became a "monopoly" because their system was not standard in any way, so they easily could push up their prices to the skies (I can assure you that a system integrator from India would do that as well immediately as soon as they can). As a result customers started to push this company to productize its solutions, which the company did. So as a result this software vendor started to be like a mini-SAP. Isn't it strange? History repeats itself.

My prediction is that the opposite will happen: integrators will suffer from being much easier to compose custom business processes based on SOA-based business process platforms (therefore integration projects being simpler, quicker and definitely cheaper), and large ERP vendors will flourish as never before.

Come back in 5 years and decide who was right

Hype and Gartner

Peter Csontos — Tue, 15 Aug 2006 11:18:00 GMT

As it happens regularly, Gartner Group has published recently a list of what's hype in their opinion nowadays. There's nothing surprising in this because hype is their business.
Not too much surprise is in there, because the majority of the list contains stuff that's the re-invention of something that's been existing for long time. For example, Event Driven Architecture is in my eyes exactly what Message Oriented Middleware is about, like as you all know, today's other popular architecture, SOA has also not been a radical invention, but rather a HTTPized, microsoftized, commoditized, derogated CORBA (OK, I know it's more than this and it is really something that's useful but if you are honest to yourself you should admit that actually that's it).
However, there's something in the list that is worth to spend some words: semantic markup languages. In other words, the Semantic Web. Since the famous article of Tim Berners Lee in 2003, it's been popping up from time to time, because some people (maybe AI folks or the ones who find it sounding interesting or who knows) push it. However, although more than 3 years have elapsed, actually no-one uses it and we can see no real application. You may be asking why. I can tell you the answer: this stuff has no use. The web is free and unstructured by definition and by its nature and whoever wants to systematize it (you can call it catalog or semantics or whatever) will definitely fail. That's why the web won over Gopher (if you remember the good old times) or why Google won over Yahoo!. I hardly can understand why Gartner believes it will "reach mainstream in five to 10 years".
That's a bubble. Probably that's where trend ends and hype starts.
Here's the link.

My experiment with Ruby on Rails - Vol. I

Peter Csontos — Mon, 17 Jul 2006 13:41:00 GMT

Yesterday evening a homepage idea came to my mind (what kinda homepage and why type of questions are not to be answered now because of being top secret ), which is a bit different from what usual content management systems like postnuke, mambo etc. or blogging / website editor stuff offers, so I started to think about how to do that. Recently I've read about Ruby on Rails being simple but relatively powerful tool for creating database-driven websites, I decided to start experimenting. Being on vacation just from today because of the kindergarten break of my children gave me some free time so I didn't hesitate and started researching immediately.

The first thing I need to learn of course is the language Ruby. I found a good tutorial here: http://www.math.umd.edu/~dcarrera/ruby/0.3/ I've almost completely gone through it, and based on that Ruby seems to be a simple but powerful, ultra-liberal, half-Smalltalk, half-Perl radically object-oriented, loosely typed scripting language.

I like it, so I keep going.

Additional documentation about the language: http://www.ruby-doc.org/docs/

(to be continued)

The analogy of the day: business software like a car

Peter Csontos — Thu, 29 Jun 2006 11:50:00 GMT

I've just read the following in a wired.com article: "CEOs like their business apps to be just like their cars: big, fast, and German.". Of course it's about SAP. I cannot say anything else about it than "God keep their good habit!"

Here's the link: The Wired 40

LAMP and SOA

Peter Csontos — Fri, 23 Jun 2006 12:47:00 GMT

Recently there's been more and more activities out there regarding why and how to use the LAMP stack connected to enterprise business applications. As recently I read two articles (this one and this one) concerning how SAP and its competitors approach this topic, I started thinking about whether today's state of the art architecture, namely SOA or more precisely how SAP applies it, enterprise SOA is a competitor of LAMP and especially scripting languages and frameworks (Perl, PHP, Python, Ruby & Rails, Groovy & Grails etc.) or they are rather complementary. My conclusion is that both approaches can be valid but the latter one seems to be more viable in the long run.

Why am I telling that?

As it can be seen from the articles mentioned, SAP is supporting LAMP much less than others. And it has a good reason. While for example PHP's history is full of being being used for implementing (mostly small, partly large) websites with some business logic but almost no so-called backend functionality, ERP and other enterprise applications of SAP are run mostly by huge or at least big enough companies.

If you consider old versions of ERP systems or smaller customers that don't want to invest in the Java stack of Netweaver, the only two ways of extensibility are changing ABAP code or use any language that has RFC connector and call BAPIs or whatever. Having this situation, in most of the cases it's not only feasible but maybe one of the best options is to use Scripting in a Box, choose your favorite scripting language and go ahead.

However, if you have to implement something for a bigger company and/or on top of a newer version of MySAP that includes Java stack, probably it's better to utilize the capabilities that SAP provides. I'm telling this because of two main reasons: one is that Java as a platform is in my opinion much more designed to be the base for applications having a relatively high architectural complexity; and two is that SAP delivers various modeling and development tools and runtime environments (for example Enterprise Portal, Composite Application Framework including Guided Procedures, Visual Composer) that make it easier and more cost-efficient to produce custom applications on top of the MySAP suite (or later on, the Enterprise Service Repository).

Last but not least, don't misunderstand me: as a skilled programmer I like very much scripting languages, mostly Perl, sometimes I don't even understand why operating systems have other user interface than bash or ksh , but the trend, that probably none of us can turn back, is to empower more and more people with less and less technology skills to implement or adjust business processes implemented with IT systems quicker and easier.

However, I don't fear that we as programmers will lose our jobs in the next 300 years or so

Bill - you know which Bill - retires

Peter Csontos — Fri, 16 Jun 2006 13:18:00 GMT

Mr. Gates announced yesterday he does something in 2008 that he says it not retirement but in fact it is.
Why would someone who's just 50 do that? You could say, he has enough money, why to work further? But in this case why didn't he retire 5 or 10 years ago? Probably he is kind of a guy who works not for money but from passion. Works until it makes sense to work.
Probably now has come the time when it doesn't make sense for him to work for Microsoft anymore.
It questions the future of Microsoft in general. They cannot grow anymore, they're going to loose being a monopoly not too much soon, and because they are not better than others at all, not being a monopoly will result in ceasing the justification of their existence.
See the articles below for more details:
Yahoo article
USA Today article

The big boss talks about unusual things

Peter Csontos — Tue, 13 Jun 2006 06:58:00 GMT

Shai Agassi, the boss of the boss of the boss of the boss of my boss (member of the board of SAP AG) was recently interviewed by San Francisco Chronicle (as he lives and works in the Silicon Valley - San Francisco Bay area, California).
Usually he talks in an official manner and about strictly business related topics, but in this article we can read about what he thinks about the future of software industry, globalization in general and even about the future of the country he was born in: Israel. It's quite interesting to read.

Welcome!

Peter Csontos — Thu, 08 Jun 2006 13:26:00 GMT

Dear all,

Hereby I'm opening my newest blog.
The reason why I'm discontinuing my old English blog (mcmlxxv.blogspot.com) is that I haven't updated it for a while anyway, and having a brand new own domain, I decided to restart it here.
I'm still updating my Hungarian blog (mcmlxxv.freeblog.hu), because it's linked from a number of other sites.
I hope I'm gonna be able to write here often, so drop by frequently!

Peter