Dan Sholler's Musings

Links for 2012-01-08 [del.icio.us]

2012-01-09T00:00:00-08:00

SAP: loathe it or ignore it, you can't like it

Links for 2011-12-11 [del.icio.us]

2011-12-12T00:00:00-08:00

SAP Community Network Blogs

Links for 2011-12-07 [del.icio.us]

2011-12-08T00:00:00-08:00

Hana prototype in a bocx
This is a service that offers to build a fixed cost HANA prototype to prove business value. ahead of the investment in HANA box
2-7381509055584468253-13419ad232d.asx (video/x-ms-asf Object)
webinar on amazon app architecture

Links for 2011-11-05 [del.icio.us]

2011-11-06T00:00:00-07:00

032165.pdf (application/pdf Object)
My snowblower manual.

Links for 2011-10-14 [del.icio.us]

2011-10-15T00:00:00-07:00

GRC-209 - Pimp My Risk Model - Getting Resilient in a Complex World.pdf (application/pdf Object)
Discussion of risk analysis in an interconnected world

Links for 2011-10-12 [del.icio.us]

2011-10-13T00:00:00-07:00

Stevey's Google Platforms Rant

Links for 2011-08-15 [del.icio.us]

2011-08-16T00:00:00-07:00

http://www.publicbroadcasting.net/wrni/news.newsmain/article/0/0/1839922/news/Quality.of.Life
My friend Joanne on WRNI

Should kids have to blog in school?

2008-09-18T18:08:00.002-04:00

I was discussing this with some friends the other day, and realized that I had a rather conflicted attitude toward this idea. On the one hand, it is quite reasonable to want to get kids in school to write, and while keeping a journal is not entirely out of fashion, the up-to-date way to do that is to write it in a blog. On the other hand, asking a 3rd grader to write to a blog that will be publicly viewable on the internet is a bit unnerving. After all, developing his or her own web presence and persona seems to be a bit ambitious, and it is unlikely that elementary school kids have the experience to temper what they write (it is clear that high-school students don't given the number of scandals (sorry too many references to be comprehensive)).

I don't know about you, but I know I think very carefully about the things I write for public consumption, whether they are the research reports that I do for my work, entries in this blog, comments on other people's blogs, statements about my current activities in facebook or twitter, and even in email. The knowledge that I have about what is appropriate to put in these kinds of communications comes from years of experience. I am troubled by the idea that we would ask an 8 year old to do the same things. On the other hand, we do ask our schools to teach our children modern, up-to-date skills. Presumably anyone reading this post would think that blogging was among those skills.

Of course, since the particular case would be a school assignment, presumably the teacher would provide some oversight. On the other hand, it seems like you are relying on the teacher to preserve your privacy in a way that is above and beyond what could normally be expected. After all, you would probably expect a teacher not to publish directly attributable statements made by your child on his or her own blog. But blanket rules are easy to follow. .

One obvious answer is to make the blog private. This way, the only readers are those within the allowed circle. Even this is unclear.. as we do not know all the parents in the class, and while our school system has an "appropriate use" contract for the internet that the students must agree to, the parents are certainly not bound by it. Even in this model, instead of just one teacher being accountable for privacy, we now have some 42 parents, plus whoever they (or the kids) give the password to.

It is also unclear to me whether or not this is "fair". Clearly, this activity creates an unfair situation for those that do not have or cannot afford a computer at home, even though both the elementary schools and the public library in our community are well supplied with computers, so there are plenty of out-of-house possibilities for access.

I am also concerned about the publicity of it all. From being a parent, it is very clear that there are elements of a child's personality that seem to be present from a very young age. There are quite a lot of children who would be thrilled by the idea of something that they have done being publicly visible, but there are just as many that would be uncomfortable with this idea. Giving a journal to a teacher whom they trust is quite a bit different than making it available to all of their schoolmates, parents, or any random person.

So, do we ask our kids to blog in school? Or do we we still keep a journal in an old-fashioned notebook? At what age is it OK to blog?

Gartner Blog Network is up

2008-09-12T21:16:00.001-04:00

Just to let you know, the Gartner Blog network is up, so you can find my posts about the technology world, and lots of stuff by my colleagues over there. I will try to dual post on here as well, but at the moment that is not working. My first post in the other blog is here.

Back in the blogosphere...

2008-09-11T17:15:00.001-04:00

Well,
Now that Gartner has changed a policy that made it tough for analysts to blog, I should be able to get back to commenting with some regularity. I am spending most of my work time these days understanding how people are leveraging SOA, and digging into specific styles like the use of Web Oriented Architecture.

We have gathered a lot of really good data about SOA adoption, justification, value capture, technology, skills from a worldwide pool over the summer, and I am spending a lot of my time analyzing and writing that up. The rest of the time is spent looking at how people are doing things, and in particular how they are following the WOA style. If anyone has any good examples they want to share, please let me know.

it's nice to be back

Bob, Dave and the curse of email

2006-05-03T11:47:00.000-04:00

Modest thread started by Bob Sutor discussing sharing documents. Dave Berlind picks this up with an good description of all sorts of alternative approaches to emailing things around. Unfortunately, the fact of life at most organizations is that email remains the primary mechanism for information sharing. While Dave did not say this directly, a lot of web junkies assume that much of the new web-facing technology will supplant email as the mechanism for sharing. Being the contrarian that I am, and just looking at how I do things, I think that the opposite is likely to happen. Instead of Wikis and blogs replacing email, we will add wikiesque and bloglike capabilities to our email systems.

Today, I rarely detach things from email. I use the Yahoo version of the X1 search engine to search my local email, and (unfortunately) have a separate search engine for my corporate intranet. This is not perfect, but I find that the distinctions that we make between thing based on their technology often do not make sense. It is very common for someone to send me something with the assumption that all I want to do is read it, and I later edit it into something different to use in another context. My entire approach is centered on email as the principle repository of information that is relevant to me.

It is a bit of a chicken and egg problem to figure out why this is since email is the primary sharing mechanism reinforces the use of email for information organizations and sharing. However, I think the more fundamental reason for this is that email is organized the way I think. When I am looking for something, aside from the topic, or some key words that might have been used, the things I am most likely to remember are when I first saw it, and who it was from. Maybe this is just because I am so tied to email, but I recall a study that was done in the (paper-based) 1980s, on how people with incredibly messy desks still managed to find things almost as fast as the compulsively neat. The conclusion was that the mess was actually organized, primarily by time and sender, as opposed to the neat people who had some kind of taxonomy they used. The neat people were better at finding one-off, obscure things, but the messy people were just as good at finding those things that related most closely to their day to day jobs. (if anyone has a reference to the study I am thinking of.. it came out in about 1987, and I remember seeing it referenced by some folks who were proposing alternate models to the relational database... see? )

So in the end, I think what has to happen is that we need to add these web-like features to email, so we can link back to shared originals, so that I can update an attachment an have the update be reflected back in the original document etc... We have already progressed somewhat in making email searchable, although this can be improved as well. Clearly, there are some problems with email (being pushed information that I am unlikely to use) but this is more of a mechanical question (is the email I am looking at a single copy for me, or a shared copy for my distribution list? )

Maybe the socialization of blogs and Wikis will replace the email organizing mindset, but I think it is more likely that they will morph to become part of the email of the future, rather than supplanting it.

UI terminology gets ugly

2006-04-18T21:12:00.001-04:00

Client/SOA - Wikipedia, the free encyclopedia

Oy, talk about buzzword proliferation..

Isn't this the same as a rich and thin client?

It has become quite amazing how confusing the discussion of client-side technology has become. IMHO, there are three issues that need to be discussed with regard to the client:

The technique used for managing session state
The page refresh model

Visualization mechanism

I am probably not dividing this up exactly right, but it seems to me that people tend to get these things all mixed up in their discussions of UIs. People assume, for example, that AJAX (which is really at its core a discussion of the page refresh model) is tightly bound to a whole host of pre-built widgets that exist in a framework (which is what most of hte AJAX frameworks provide). However, the AJAX-ness of the thing is not actually int the framework, but in the fact that framework elements and their data can be loaded on demand within the page. This Client/SOA word (ugh!) is really about managing session state on the client, which is hardly a new practice (That is why you have all those cookies, after all.. ). This type of interaction is necessary, but not sufficient for a REST-style architecture.

I hope that the market reaches some rapid consensus on terminology here.. this confuses me, and I get paid to pay attention to this sort of stuff.

Sorry this blog has been silent

2005-11-21T16:21:00.000-05:00

Sorry this blog has been pretty silent. I am sure there are no readers left :)

However, I have a good excuse. We had a new family member arrive about a month ago, and having two is definitely more complicated than just 1+1.

Will get back to it eventually...

Dan

The relationship between SOA and Ajax

2005-09-09T17:33:00.000-04:00

Several discussions about the relationship between SOA and Ajax are floating around now. (First one was this entry from Dion Hinchecliffe, with a dissenting opinion by Nick Malik, and some additional commentary by Dare Obasanjo.

This is actually the most interesting part of the entire discussion of the evolution of web services. It is very clear that there will be a large set of fairly fine-grained and specific services that will be made available due to the creation of AJAX style applications. It is also true that most of these services will not be terribly reusable. However, in parallel, there are a large number of initiatives that are attempting to create non-browser UI mechanisms (see Konfabulator. These mechanisms would be perfect candidates to consume the services that are designed for Ajax. (Yet another example of the changing nature of the contract between server and client, ) This kind of thing represents a specific set of services that will be implemented in a particular style, and will not use SOAP and WS-* stuff. There will be other sets, one of which are those connections that are most similar to traditional integrations, which will mostly use SOAP. There will be several other categories as well.

The point is that what distinguishes these services (and why they will probably use different mechanisms) is the expectation that is needed from the infrastructure. The AJAX style services need to be fast, and also need to be relatively easy to program. The Integration services need what they have always needed.. Message oriented middleware.. which is a collection of capabilities (security, reliability, state transfer, etc... ) which are most likely to be implemented with SOAP. I expect that many of the content related services will be ATOM, RSS or something very like both.

So on the one hand, the idea that AJAX will be a manifestation of services as many people define them is not true, they are a slightly different class of beast. However, they will be a very important class, and this phenomenon will have a tremendous impact on the concepts and use of services on the web.

[Furl] Daniel Sholler's Headlines for August 23, 2005

2005-08-23T05:51:00.000-04:00

Daily Furl Headlines for August 23, 2005
From Daniel Sholler - http://www.furl.net/members/dsholler

Envinsa Location Platform - Features
Rated 3 in topic web services
From extranet.mapinfo.com on August 22 at 2:30 PM
<http://www.furl.net/item.jsp?id=4262053>

MasterCard Cashes in on Web Services
Rated 3 in topic web services
From eweek.com on August 22 at 2:29 PM
<http://www.furl.net/item.jsp?id=4262044>

________________________________________________________________________
Save and share anything you find online - Furl @ http://www.furl.net

[Furl] Daniel Sholler's Headlines for August 20, 2005

2005-08-20T05:00:00.000-04:00

Daily Furl Headlines for August 20, 2005
From Daniel Sholler - http://www.furl.net/members/dsholler

Robert Lefkowitz - The Semasiology of Open Source
Rated 3 in topic Open Source - E-business
From itconversations.com on August 19 at 8:59 PM
<http://www.furl.net/item.jsp?id=4235242>

IT Conversations: Werner Vogels - E-Commerce at Interplanetary Scale
Rated 3 in topic Technology
From itconversations.com on August 19 at 8:58 PM
<http://www.furl.net/item.jsp?id=4235232>

RSS Mystifies Most Blog Readers
Rated 3 in topic Standards
From editorandpublisher.com on August 19 at 8:47 PM
<http://www.furl.net/item.jsp?id=4235170>

Cover Pages: Initiatives Ramp Up Work on XML Naming and Design
Rules Specifications.
Rated 3 in topic Standards
From xml.coverpages.org on August 19 at 8:21 PM
<http://www.furl.net/item.jsp?id=4235033>

Cover Pages: Business Narrative Markup Language (BNML) Proposed
for eContracts.
Rated 3 in topic Standards
From xml.coverpages.org on August 19 at 8:20 PM
<http://www.furl.net/item.jsp?id=4235022>

tecosystems: Zimbra & the Messaging Opportunity
Rated 3 in topic Open Source - E-business
From redmonk.com on August 19 at 7:57 PM
<http://www.furl.net/item.jsp?id=4234914>

________________________________________________________________________
Save and share anything you find online - Furl @ http://www.furl.net

Small business giveaways?

2005-05-23T22:26:00.000-04:00

At the risk of talking about something that everyone knows already, I have just started using picasa, which is a great relief after prior attempts to struggle through other editing programs. It does just about everything I want, and does it quickly and simply. Another good example of useful stuff that is available for free on the web, although I would be curious to know how many people actually do the things that will make Google any money. Unfortunately, the genie is out of the bottle, and if you want to attract attention on the web, (or at least, if you want to attract a broad audience) you need to give them something, and the quality bar on that giveaway is going up all the time. I wonder how long before something of the complexity of Microsoft Money is given away? Not long, I suspect.

The more interesting question is whether this effect bleeds over into the small business software space. After all, this is a market that does share many attributes with various consumer segments. In addition, there are plenty of business services companies that would underwrite some pretty sophisticated software in order to gain market share in this segment (ADP , Experian, Cintas, and Iron Mountain , are a few names that spring to mind). This is something that these companies do to some extent already, but mostly tied to the purchase of the services. I believe that they will eventually give software away that has value in its own right, absent those services, the same way that Google gives picasa away.

I think this will create and opportunity for software developers who focus on the small business space. If this goes well, it should increase the quality and amount of software that is available to small businesses. The downside of course is if these things are all implemented in ways that make them difficult to integrate, that could be an annoyance. However, in the small business context, much of that integration could be accomplished with basic mechanisms (XML documents written to files and uploaded) without causing tremendous business pain.

It is curious that everyone focuses on open development and open source as the potential source of free software for businesses, when in fact it is also possible that this software will come from business service providers who want you to use their services, and where developers will get paid for their work. It also suggests an alternate business strategy for an organization like Compiere.

James Governor's MonkChips: End of Some Analysts. Join The Dots with Depth in Pockets

2005-05-23T13:36:00.000-04:00

James Governor's MonkChips: End of Some Analysts. Join The Dots with Depth in Pockets

James, as usual, has some good observations here. However, I think in this case his emphasis is not quite in the right place. It is not just about blogs and RSS. There is no question that an energetic individual could in fact navigate his or her way around and discover a variety of trusted sources of information from practitioners, and use that as one of the inputs for guiding their decision making. The challenge at this point is that for most folks, those sources of information are too numerous, and their quality and points of view too difficult to discover without reading all the background info (as opposed to just looking at the entry on the subject of interest at the moment. ) I believe that there will continue to be value in people who aggregate this informed opinion, supplement it with their own research and observations, and offer to distribute that for a fee. (Please note: I am quite biased here, being one of those "industry analysts" he is writing about).

I do agree that the traditional publication model is not as valuable as it once was, since there are forms of communication that people are coming to expect (with Blogs, etc. ) that are much more interactive and ephemeral than the published research report that is expected to stand for a year or more. It is quite clear that everyone in this industry needs to be able to communicate in the fashion demanded by their (our) clients, and that means being much more rapid and participatory. However, I also believe that there still is value in some of the traditional research forms, because there are situations that require long term analysis, and where long term trends can be identified and published in a way that makes them useful. The trick is going to be to figure out how to retain the depth of expertise associated with the specialist model, and layer over that a more generalist integrated viewpoint.

Of course, one disruptive influence on this would be a reasonable and trusted way to actually search and evaluate content that is available on the web. This may happen through some of the folksonomies that are being developed (e.g Furl, del.icio.us, etc. ) However, at the moment, these suffer from the same trust issues that the general web content does (I subscribe to individual's Furl bookmarks, because I know they are people who are interested in some of the same things I am).

At the end of the day, even that is not enough, since what really has to happen is that this analysis has to make a difference. Sometimes it is possible to make that difference through a written or otherwise published (podcast, etc. ) means, but in many cases that means that you need to actually talk to the people involved, understand what it is that they are trying to accomplish, what their constraints are, and then help them to come up with a plan based on the analysis of experiences of others, and analogous situations. This is why there will be a consultative component to all of this industry analysis, and that component will grow. This may also challenge an organization like RedMonk as they try to grow their business and improve their margins at the same time. However, James and his colleagues have picked the right challenge.

developerWorks : Blogs : Tom Glover

2005-05-23T12:20:00.000-04:00

developerWorks : Blogs : Tom Glover

IBM is doing a good job trying to kickstart the notion of profiles. This is an interesting commentary on WSI, and reinforces the perception that WSI is moving too slowly, and has become a validation on things that are already fully baked. I believe that WSI must adopt a more forward looking view such as IBM is proposing with these profiles, or they will be doomed to irrelevance.

These profiles are interesting in that they are trying to sort out what it is that various clients in various industry segments actually need. However, of course IBM is also doing this to reinforce their leadership in the web services standards domain. Hopefully they will actively promote these profiles as things that anyone can use, as opposed to treating them as IP that IGS can take advantage of. I believe that is the intent, since even though they would be open, IBM as the progenitor would have an advantage with a customer looking to engage someone to help them implement these things. The real interesting part will be when we start to see other companies providing support for these things both in tooling, and as service offerings.

Dave Orchard's Blog: Partial Understanding

2005-05-18T09:06:00.000-04:00

Dave Orchard's Blog: Partial Understanding

As usual, Dave Orchard sums up the technology issues really well. The idea of having partial schemas, and a means of combining them is definitely on the right track. Ideally, we need to have something that supports separation of concerns. There is then the issue of specifying to whom the schema subset applies. I believe that this must be outside the schema language, and not necessarily predetermined by the publisher unless that is a specific constraint (only somatic can read this part).

updated 5/25/05 to fix typos

Tips on how to blog safely - Apr. 8, 2005

2005-05-16T21:18:00.000-04:00

Tips on how to blog safely - Apr. 8, 2005

OK, I am sure I am not the only one who saw this as a result of looking at IBM's blogging policy. If this article had been written even a year ago, when blogging was not quite as mainstream as it is today, I would have not been at all surprised. However, it really does strike me as odd that a reasonably mainstream media house could be this clueless about the one thing that has shaken up mainstream media in the last few years. Don't they remember that Matt Drudge scooped them on the Monica scandal? what about the infamous white house blogger?

It just astounds me how blind organizations can be when staring change in the face. Clearly, no individual in the blogsphere will become as institutionalized as the major media outlets, but those outlets do need to realize that there will be some folks that join them on the front end of the power-law curve. This is true in every industry that relies on content and content distribution. It is no longer enough to be the disseminator of information, you need to do something more. The (IMHO regrettable) rise of Fox News occurred precisely because someone over there realized that they needed to offer more than just good reporting. They chose to offer opinion and ideology mixed in, which thus far has seemed to be a good selling point.

Just as I would not take advice about much to drink from a bartender, it is clear you should not take advice about blogging from mainstream media sources.

Common Craft - Social Design for the Web: Wiki This- A Model for Customer Support Using Blogs and Wikis

2005-05-13T13:47:00.000-04:00

Common Craft - Social Design for the Web: Wiki This- A Model for Customer Support Using Blogs and Wikis

Definitely on the right track here. This model of enabling web-based communication with customers is just now penetrating the IT industries, and I believe it will become commonplace for many companies in a very short period of time.

There are actually two problems mentioned in this article, which probably require two distinct models to solve them. One is the problem of presenting information to an audience according to an ordered taxonomy that is predetermined. The other is allowing the user to create order out of the volume of information that is posted. The first is accomplished through the wikiThis notion, which allows the generators of blogs (and the reader of comments) to organize information for future reference according to a predetermined taxonomy (or to build out that taxonomy). Unfortunately, in many cases the taxonomy gets in the way of actually using the information. For example, I would classify this article as part of my set of information about how creating communities is a critical part of business in the information age. Others, who have different viewpoints, and have an interest in, say, blogging as a social phenomenon would classify this in a completely different way (where the blogging aspect, not the community aspect is the most important). Even if I knew that this other person had classified this article in their taxonomy, it would probably take me a long time to find it, if I managed it at all. I don't know about the rest of you, but I often have trouble finding things in the Microsoft help system, for example. This is not because it is a bad system, but just because no matter how well organized they make it, the organizing principle will be different from the way I view and organize the world.

Therefore, another critical aspect of this is enabling the readers to create custom taxonomist. This by itself does not help anything (or at least, not much) since the custom taxonomy is understandable only to the individual. The other part is that people must be able to subscribe to the taxonomies of trusted individuals, so they can use the time that those individuals take to identify items of interest to help organize their own information. Today, I use Furl in this way, and I expect that once the information gets to a certain level of complexity, companies will have to support this idea as well. This is the main problem with the current Wiki notion, in that there is no way to segment the organization that I see, other than with fairly primitive tools like watchlists. I also expect that companies will have to add a new step to the process described in this article, that creates new organizations of the information based either the organizations of leading consumers of that information, and also based on the companies idea of constituencies for information.

But in general, the wiki does not let people organize information for themselves, nor does it allow them to refer to a previous organization (think of it as a search query). Today, I use a combination of Furl, my RSS reader settings, and my browser bookmarks to organize the internet for me. The fact that these are multiple tools causes me endless headaches, and unifying all of this into a single source will also be important for making it convenient. This is a bit of a trick, since on the one hand you want a single set of tools and mechanisms for organizing the information, and on the other hand you want it to be flexible enough to handle the interesting things that people do with it. Flickr has been a great example of something that is full featured enough to be usable, and simple enough for lots of folks to use. I am waiting to see what happens with this wiki/blog/tagging combination described here. It seems like a great business opportunity of someone can put it together for general business use.

2005-05-13T10:17:00.000-04:00

http://www-1.ibm.com/press/PressServletForm.wss?MenuChoice=all&TemplateName=ShowToPrint&SelectString=t1.docunid=7658&TableName=DataheadApplicationClass&SESSIONKEY=any&WindowTitle=Press+Release&STATUS=publish&ShowContacts=$ShowContacts$

An interesting, but hardly unexpected development. The question was not whether IBM would get into Open Source J2EE, but when. I am surprised at the timing of this (it is about a year or so earlier than I would have predicted) but I assume the opportunity was timely.

While lots of folks will make a big deal out of this, it is just a sustaining event. After all, IBM has been a big supporter of Apache in the past, They have already begun the conversation around open source and Java, and have made commitments to other open source initiatives (Eclipse). There are three major things that this announcement does: Makes it clear IBM is backing Geronimo, gives IBM a chance to experiment with the Support only licensing model, and puts pressure on both JBoss, Jonas, and BEA. It is the last bit that is curious, since in the short term the legitimizing effect of IBM endorsing an open source J2EE container will probably actually help JBoss. On the other hand, I do not believe IBM sees JBoss as a long term issue. (After all, the primary reason for building technology is so that they can have more credibility and expertise in that technology in their service organization.. ) The pressure on BEA is the continued devaluation of the core J2EE container.

At the end of the day, this is just an embrace of the inevitable. It also shows that scale can create opportunities for vendors, since IBM's scale allows it to embrace these different strategies, whereas BEA is challenged to take such an approach.

I believe BEA needs to get much more aggressive here. They are the ones that have the most to lose from open source alternatives (after all, they have the position right now of being the neutral technology, and therefore get the endorsement of the various systems integrators).

I also believe that the only way that they could differentiate would be to deviate from the J2EE party line. A server that natively supports and enhances things like spring and hibernate would be very interesting. Also a large scale commitment to AOP would help as well.

The real SOA debate

2005-05-13T09:57:00.000-04:00

The real SOA debate

Most of the confusion around SOA is about whether or not SOAP or WSDL is the central construct in a service oriented architecture. This is a difficult debate to resolve, because the answer goes against most people's well trained instincts.

I don't believe that anyone disagrees with the idea that the programming model is critical to the success of SOA, or that at the bottom layer there will be a multiple protocols. I think it is a question of what will vary over time. In my opinion, a requirement for SOA is to have a universal messaging network with intermediary processing. In the past, this was achievable with proprietary technologies, since the domain that was described as "universal" was a subset of the corporate intranet. However, for today (and the future) that domain must be expanded to be the internet and the relationships in the value chain. The thing that enables organizations to treat the whole thing (internet/intranet) as a coherent network is the SOAP protocol. This protocol will overlay the messaging network, the same way that IP overlayed the bit-level network. (sorry, back to the IP analogy again.. ) . At the time that IP was introduced, there were many segments of the networks that were not IP based, and there were many mechanism that were used to enable organizations to bridge those segments. After a while, when most applications were using the IP based communications protocols, many of those segments were replaced with native IP-based segments. The same thing will likely happen with the SOAP network, in that today there are many messaging protocols that will be used, but over time some of those will be replaced with native SOAP (or standards-based SOAP over HTTP).

To continue the analogy, one of the things that was critical for the success of IP was Berkeley sockets and winsock. These made it very simple to program the universal network overlay (as opposed to having to program to a protocol specific network). Using these things in applications was valuable, because then the applications did not have to worry about the specific protocol, and that underlying protocol could be varied based on intermediaries in the network. (as opposed to being bound to a specific protocol at the endpoint). The same analogy exists with SOAP and WSDL, in that a model where applications write to a SOAP interface (whatever that is) and then bind that interface to the SOAP overlay (as opposed to the specific messaging transport layer) then allows the variations of message transport to take place independently of the applications.

The concern that I have about the focus on WSDL as the defining element of SOA is that it leaves out the other part of the equation. If you can write programs to WSDL< but the endpoints are then hardwired to their protocol specific WSDL implementation, that will mean that the network is not nearly as flexible, and folks that do it this way will eventually run into change problems. This will make it very difficult to change messaging protocols (need to change them at the endpoints). Even with some kind of dynamic protocol selection in the stack at the endpoint, you still lose the ability to insert intermediaries in the network that can act on behalf of multiple endpoints (because the intermediary will be required to deal with the details of the individual messaging protocol, as opposed the standardized SOAP structures.

If you buy this premise, then the fundamental question is whether WSDL is the appropriate endpoint binding language for the universal SOAP-based network, or whether it needs to be improved. There is also the question of what exactly is defined by the interface definition, in the sense that there is far more metadata required to actually use a service in a mission critical context than is just captured in MEPs and XML schemas. (the stuff that was begun in WS-Metadata, for example). IMHO, the next "unix wars" will be the struggle by various companies to define this metadata bundle. While this will be a challenge for organizations implementing SOA, it does not break the fundamental

Jonathan Schwartz's Weblog

2005-04-29T16:26:00.000-04:00

Jonathan Schwartz's Weblog

I find contains blog interesting, although I often find myself disagreeing, or agreeing with his conclusions and disagreeing with the reasoning. In this entry he puts in a plug for SunRay as an ecologically responsible alternative to PCs. I think that is correct, but I think that much of Sun's thinking around SunRay misses the point (or at least, what I think is the point). The trick to making SunRay successful in this day and age is to create a standard (ie. device independent) metadata format. The cards that people use to configure SunRay are the critical architectural point. After all, there will be a SunRay 2 sometime (probably soon) and they need to have compatibility with the configurations that users have already created (I have no idea who uses SunRay at the moment). This argues that the things Sun needs to work on is that metadata model, and opening it up as much as possible. this would then give people incentive to buy these kinds of devices (since they would not then be locked into particular hardware, and also because the utility of that metadata could be more widespread (have it configure my phone, for example) ) but at the same time it would give Sun a chance to innovate with the actual desktop device.

This is a bit more complex than it appears of course, because it is not only a matter of standardizing metadata, but of standardizing the services to which the metadata refers. However, this seems achievable, and if Jonathan's statements about Sun's use of open licenses can be believed, it would fit with the strategy. We'll see if it happens. This would then open up a really interesting alternative computing model, that was independent of the Windows desktop (which I assume is what Sun wants... )

SAP - SAP and Macromedia Advance Usability of Enterprise Software

2005-04-26T10:07:00.000-04:00

SAP announced a partnership with Macromedia today that I believe will be the bellwhether for Rich Interactive Applications. Today, most of this interactivity is in brochureware, or in catchy games and other gimmicks that are attractive, but are secondary to the business capabilities of the applications. If the SAP user base (an inherently conservative bunch, for the most part) starts to actually make use of FLEX that will indicate to me that RIA is ready for mass adoption. My personal bet is that this will not happen, but I would like to be proven wrong.

SSDL Overview

2005-04-25T09:26:00.000-04:00

SSDL Overview

I like this a lot, although it seems to me that there should be some way to reconcile the depictions of MEPs with BPEL.. does anyone know of an effort like this under way?

The reason this is so compelling is that it concentrates on the basics, the protocol, as opposed to the programming language binding (a la WSDL). In the context of Service Oriented Architecture, programming models don't matter. (Of course they do matter, quite a lot, but they do not matter to the architecture ) Because WSDL is pretty much unrelated to the underlying XML protocol (SOAP) used for the service, it allows people to fall back into bad habits.. in terms of thinking about the programming model as the critical design point of the architecture, not the protocol. The big "aha!" with SOA is that systems are networks, and the thing that binds them together is a common network protocol, not a common programming model. Whatever interface language is used (SSDL, WSDL, something else), the key consideration must be that it be linked to a common protocol, and today, that protocol is SOAP.

Comments on Geoffrey Moore's presentation

2005-04-19T18:09:00.000-04:00

I just had the opportunity to see Geoffrey Moore's presentation on Core and Context. He has done this presentation in a number of technology venues.

There are a few interesting things that he points out that are really useful, and one big thing that I think is not quite right (or at least, not quite fully formed).

One of the more interesting ideas is about different people's roles and skills in the organization. Traditionally, there has been a dichotomy between innovators and operators. There are folks who create new stuff, and there are folks who make stuff work at scale. The thing that Moore added to this, which I believe is a critical insight, is that there are a class of people whose job it is to deploy. Deployment activities are about taking innovations and migrating them to mission critical endeavors (I think of this in the context of IT projects, but of course this applies to any business activity. ) This separation, and the idea that we should not expect people whose primary role and skill set is around innovations, or around operation and optimization to be good at deployment is extremely important. From an IT perspective, this separation is often given lip service in the "plan-build-run" cycle, but really thinking about staffing in these terms is relatively unusual.

Unfortunately, I think he missed on big point.. (and I need to give credit to Nick Gall for discussing this with me.. We both had the same reaction to this point)

Moore presents a series of steps about what has to happen to prevent organizations from becoming sclerotic. In particular, he points out that what happens is many of the activities (and the IT systems that support them) begin as differentiating activities (in his terminology, "core"). They then typically get implemented at scale so the company can gain the benefit of that differentiation, in other words, they become mission critical activities. Unfortunately, the differentiation erodes over time, and a some point these activities become undifferentiated (i.e. do not command an economic premium). The danger is that they remain mission critical, and organizations often continue to increase the resources that are dedicated to these activities, which means that fewer resources are available for really differentiated stuff. (this is my paraphrase here.. )

He insists (quite rightly) that organizations must actively get their processes out of this state, or they risk their ability to maintain margins and profits. He proposes a series of steps that need to happen to systems that are in this state (I am applying this to IT systems here by way of example). The steps are (my paraphrase again.. )

Centralize

Standardize

Modularize

Optimize

Outsource

The one point I think that he missed, which is one of the primary drivers for adopting service oriented architectures and agile methods is that once a system is in this state (mission critical but not differentiable) it is extremely hard to justify spending significant amounts of money on it. The first two steps (centralize and standardize) are generally viewed as money saving opportunities ,but the modularize step costs a great deal. This cost is the primary barrier to completing this process, and means that these systems often remain in this bad state indefinitely, constantly requiring increased investment just to maintain a capability that does not add value.

In order to unwind this conundrum, people must begin with a modularized approach. The only way that systems which can be built that is amenable to these steps is if at the time the system is built and deployed (i.e. when the system is still "core" or differentiable) it is done in a way that simplifies modularity. This argues that the increased modularity of systems that are built in a service oriented fashion has some significant long term benefits as these thing move through this core to context lifecycle.