All posts by : Radovan Janecek: Nothing Impersonal

1 of 108 | SOA Symposium - I had two presentations on SOA Symposium last week. One was on avoiding SOA pitfalls, such as EAI in angle brackets, lack of governance, ESB approaches, and similar. The other one was focused on various considerations you should think about when choosing between Atom (REST) and SOAP (WSA). I’ll post link to both decks as [...]...

2 of 108 | Gartner on REST - I spoke about BTO for SOA on Gartner Application Architecture, Development & Integration Summit. But instead of talking about HP's solution I tried to share key lessons learned from being an SOA vendor for over 8 years and more recently trying to apply SOA accross large BTO portfolio. I was careful not to be so [...]...

3 of 108 | Good Data Went Live - Roman is apparently making fast progress with his new team! Good Data went out from stealth mode with new web site on business intelligence. Focusing on collaboration is very interesting move as there has been a lack of this aspect in enterprise software in general. Plugging humans more deeply into enterpise software is the mainstream now with all the wikis replacing heavyweight portals, email-based workflow systems, mashups, and new kinds of groupwares. Business intelligence tools, used by people to make sense out of their data, should have been offering collaborations for ages. Best luck with fixing that! Btw, even Systinet was trying to enter this territory five years ago with collaborat......

4 of 108 | Cape Clear Acquired - So, one of the last independent SOA companies got acquired. As Roman wrote, Cape Clear used to be our competitor in Systinet times. Tough one. Workday gets great integration expertise. If they provide pre-packaged integration solution for their SaaS offering it's going to be very successful, I'm sure. And they can afford to get rid of ESB marketing buzz as it will not be relevant for their business any more. I wish good luck to all Cape Clear folks.   ...

5 of 108 | Information Modeling for SOA - Jeff misses information modeling for SOA in his post on Next Phase of SOA. I absolutely agree. Indeed, while we are talking about ESBs, BPELs, WS-*, REST, etc. - the lack of systematic approach to how we model SOA can result in mess and chaos (this time in angle brackets). That's why I was asking so many times about REST and Contract for example. Looking from BTO perspective, the issue is even wider and tougher. It is not enough to make BTO RESTful. How to agree on the model that comes from at least three different worlds: CMDB, SOA, and Data Warehouse? Is such agreement needed? Is there better and more pragmatic (loosely coupled) way of having multiple models that still make sense together......

6 of 108 | Is REST eclipsing SOAP? - Joe McKendrick asks whether REST is eclipsing SOAP as a standard of choice as we move towards SOA. Comparing REST to SOAP is very strange because one is architectural style while the other is a payload envelope. Let's ask more accurate question: Are standards like ATOM or APP eclipsing SOAP? Yes, definitely. For very simple reason - their adoption is way broader and tools support way better. Therefore if you externalize your data in Atom feed then it can be reused way more easily by other applications or humans than if you do the same in the SOAP envelope. If we add other data formats in mix such as JSON, OpenSearch, and some microformats - the 'eclipsing' question is very easy to answer. T......

7 of 108 | Chief Architect, BTO, HP - I've got very exciting opportunity to try a job of Chief Architect for BTO (Business Technology Optimization) portfolio at HP. BTO is one of the largest family of products for IT. I'm going to focus on 'external' architecture in order to allow for consistent and scalable integrations. I hope that my SOA background will be helpful.  Whoever reads my blog - keep fingers crossed please ...

8 of 108 | Atom-* - I went thru Dare's post on APP limitations. Good reading! We are using REST for many years and some of new functionality uses APP. It's too early to make conclusions but we didn't hit the wall yet. I just hope that guys at Microsoft are not going to create Atom-* that will establish reliable, secure, transactional, no-loss, hierarchical, bulk operations ready… middleware. No more RSSBuses or AtomBuses please. ESB is enough Update: I missed many posts on that. Bill: APP on the Web has failed: miserably, utterly, and completely. James:Silly. Joe: In which we narrowly save Dare from inventing his own publishing protocol. Stefan: APP not General Purpose?. Matthias: Partial APPdates. .   ...

9 of 108 | What should HP do in SOA - Dave suggests 5 Things HP Needs to do to make an Impact in the SOA Market. All resonate with me. SOA Modeling is very interesting space because, as Dave correctly points out, there is nothing appropriate on the market. I do not have any finalized opinion on this yet but SOA modeling is going to be very different. More collaborative, less rigid. As for the best practices and thought leadership - I believe in tools-enforced behavior. This is what we are focusing on in R&D. HP needs to do much more in marketing and products. ...

10 of 108 | HP SOA Launch - HP launched SOA yesterday. I’m very happy ‘our’ SOA not only survived the acquisition but actually got way stronger than ever before. Guys were working hard on the new release of Systinet2 (HP SOA Systinet now) ignoring the transition pain and acquisition madness. The new version brings many new tools, applications, platforms support, and, most importantly, it distills years of SOA governance experience (painfully gained many times). Combined (and integrated) with other SOA pieces coming from Quality (Service Test) or Management (Business Availability Center or SOA Manager) the overall SOA solution is very powerful (if not the most). You can hear HP executives talking about SOA here or her......

11 of 108 | Pat Helland is Back - After working on the other side of SOA equation, Pat is back at Microsoft. A lesson from Amazon: In this paper, I argue that real scalable systems simply do not attempt to do cross-system transactions. Let’s make this even stronger: for real scalable systems simply do not attempt to do any cross-system ‘middleware’ features - transactions, reliable delivery, transformations… But yes, transactions are number one. ...

12 of 108 | SaaS and SOA (and Google) - I went thru posts discussing SOA and SaaS [1], [2], [3], [4]. I'm kind of surprised there even is such discussion because it seems so obvious to me: you need to implement SOA in order to deliver good SaaS. Do you want to see how difficult this is? If you use personalized Google home page try to open it now. I lost about 6 months of my personal data today looking at stuff dated back to November... I don't care whether this is alpha, beta, gamma, or delta - I just need to seek for more reliable alternative to such 'service'. Update: The bug has been fixed. The fact that service provider is not willing to share root cause of the problem is not very convincing though....

13 of 108 | ESB prevents SOA - There is another thread of never-ending ESB discussion. Javier posted nice list of ESB skeptic articles. I see more and more people realize ESB is irrelevant to SOA (if not harmful). This was inevitable because ESB is conceptually EAI and not SOA. It doesn't matter whether particular ESB product is built as an extension of legacy MOM (majority of these products) or from scratch as 'pure web services' bus. At the end, it is some sort of integration server(s) - this time in angle brackets. Unfortunately, ESB vendors along with some analysts convinced a lot of people about ESB being central to successful SOA implementation. Using this Snake oil will lead to failure that sometimes will get dec......

14 of 108 | ESB prevents SOA - There is another thread of never-ending ESB discussion. Javier posted nice list of ESB skeptic articles. I see more and more people realize ESB is irrelevant to SOA (if not harmful). This was inevitable because ESB is conceptually EAI and not SOA. It doesn't matter whether particular ESB product is built as an extension of legacy MOM (majority of these products) or from scratch as 'pure web services' bus. At the end, it is some sort of integration server(s) - this time in angle brackets. Unfortunately, ESB vendors along with some analysts convinced a lot of people about ESB being central to successful SOA implementation. Using this Snake oil will lead to failure that sometimes will get d......

15 of 108 | HP Closes on Mercury Interactive Buy - So the deal is closed! This is very good news for all SOA stakeholders because the best SOA vendor just emerged! ...

16 of 108 | HP Closes on Mercury Interactive Buy - So the deal is closed! This is very good news for all SOA stakeholders because the best SOA vendor just emerged! ...

17 of 108 | TIBCO to Resell Systinet Registry - After BEA and Oracle, TIBCO has joined 'the club'. It's exciting to see how our products lead the market towards registry-centric view of SOA. Our registry still doesn't solve the world hunger though. ...

18 of 108 | TIBCO to Resell Systinet Registry - After BEA and Oracle, TIBCO has joined 'the club'. It's exciting to see how our products lead the market towards registry-centric view of SOA. Our registry still doesn't solve the world hunger though. ...

19 of 108 | SOA vs. Web 2.0: Governance or Rogues? - Mike Stevens posted his thought (via Don) about soa vs web2.0 from the governance perspective. Although the post is too short and somehow focused on wiki, the governance perspective is very interesting. It looks like Mike suggests the enterprise guys should be more friendly to the Rogues. Well, I bet most enterprise project managers will say there is a lack of governance in their IT and that there are still too many Rogues ;-) Btw, my periodic hunger for REST and Contract has similar roots I guess. ...

20 of 108 | SOA vs. Web 2.0: Governance or Rogues? - Mike Stevens posted his thought (via Don) about soa vs web2.0 from the governance perspective. Although the post is too short and somehow focused on wiki, the governance perspective is very interesting. It looks like Mike suggests the enterprise guys should be more friendly to the Rogues. Well, I bet most enterprise project managers will say there is a lack of governance in their IT and that there are still too many Rogues ;-) Btw, my periodic hunger for REST and Contract has similar roots I guess. ...

21 of 108 | On ESB - There is also interesting ESB thread on the SOA group. I guess I don't have to write down my opinion on this blog any more....

22 of 108 | On ESB - There is also interesting ESB thread on the SOA group. I guess I don't have to write down my opinion on this blog any more....

23 of 108 | On ESB - There is also interesting ESB thread on the SOA group. I guess I don't have to write down my opinion on this blog any more....

24 of 108 | Rise and Fall of CORBA, Web Services, and others - There is a very good article by Michi Henning (via Mark) on why CORBA (my ex-love) failed. It's funny that while reading this article I was agreeing with almost every sentence (except some comparisons to Web Services). However, at the end, I concluded the article missed the root cause of the adoption problem. It focuses on technical and OMG-related issues. What about the way CORBA was used? IT organization is usually organized as silos. This is very efficient from many perspectives. It's easier to assign responsibilities, measure performance, and so on. Well, and software vendors provide software that fits within this structure. The only problem is once such organization wants to adopt ser......

25 of 108 | Rise and Fall of CORBA, Web Services, and others - There is a very good article by Michi Henning (via Mark) on why CORBA (my ex-love) failed. It's funny that while reading this article I was agreeing with almost every sentence (except some comparisons to Web Services). However, at the end, I concluded the article missed the root cause of the adoption problem. It focuses on technical and OMG-related issues. What about the way CORBA was used? IT organization is usually organized as silos. This is very efficient from many perspectives. It's easier to assign responsibilities, measure performance, and so on. Well, and software vendors provide software that fits within this structure. The only problem is once such organization wants to adopt s......

26 of 108 | Rise and Fall of CORBA, Web Services, and others - There is a very good article by Michi Henning (via Mark) on why CORBA (my ex-love) failed. It's funny that while reading this article I was agreeing with almost every sentence (except some comparisons to Web Services). However, at the end, I concluded the article missed the root cause of the adoption problem. It focuses on technical and OMG-related issues. What about the way CORBA was used? IT organization is usually organized as silos. This is very efficient from many perspectives. It's easier to assign responsibilities, measure performance, and so on. Well, and software vendors provide software that fits within this structure. The only problem is once such organization wants to adopt s......

27 of 108 | Rise and Fall of CORBA, Web Services, and others - There is a very good article by Michi Henning (via Mark) on why CORBA (my ex-love) failed. It's funny that while reading this article I was agreeing with almost every sentence (except some comparisons to Web Services). However, at the end, I concluded the article missed the root cause of the adoption problem. It focuses on technical and OMG-related issues. What about the way CORBA was used? IT organization is usually organized as silos. This is very efficient from many perspectives. It's easier to assign responsibilities, measure performance, and so on. Well, and software vendors provide software that fits within this structure. The only problem is once such organization wants to adopt s......

28 of 108 | REST and Contract - I spent some time today on the SOA mailing list with my favorite REST and Contract subject. I'm copying some content to my blog: Lack of contract is an issue from enterprise development perspective: It's interesting how many REST discussions started by arbitrary topic end up at "what is the contract"? It's my personal issue as well. I do not share traditional anti HTTP complaints you hear from enterprise architects: lack of reliable messaging, lack of distributed transactions, and stuff like that. Many of these complaints have EAI (today ESB) roots and are largerly irrelevant to SOA (including HTTP/REST). However, I see lack of contract as the main issue and perhaps even showstopper in......

29 of 108 | REST and Contract - I spent some time today on the SOA mailing list with my favorite REST and Contract subject. I'm copying some content to my blog: Lack of contract is an issue from enterprise development perspective: It's interesting how many REST discussions started by arbitrary topic end up at "what is the contract"? It's my personal issue as well. I do not share traditional anti HTTP complaints you hear from enterprise architects: lack of reliable messaging, lack of distributed transactions, and stuff like that. Many of these complaints have EAI (today ESB) roots and are largerly irrelevant to SOA (including HTTP/REST). However, I see lack of contract as the main issue and perhaps even showst......

30 of 108 | REST and Contract - I spent some time today on the SOA mailing list with my favorite REST and Contract subject. I'm copying some content to my blog: Lack of contract is an issue from enterprise development perspective: It's interesting how many REST discussions started by arbitrary topic end up at "what is the contract"? It's my personal issue as well. I do not share traditional anti HTTP complaints you hear from enterprise architects: lack of reliable messaging, lack of distributed transactions, and stuff like that. Many of these complaints have EAI (today ESB) roots and are largerly irrelevant to SOA (including HTTP/REST). However, I see lack of contract as the main issue and perhaps even showst......

31 of 108 | REST and Contract - I spent some time today on the SOA mailing list with my favorite REST and Contract subject. I'm copying some content to my blog: Lack of contract is an issue from enterprise development perspective: It's interesting how many REST discussions started by arbitrary topic end up at "what is the contract"? It's my personal issue as well. I do not share traditional anti HTTP complaints you hear from enterprise architects: lack of reliable messaging, lack of distributed transactions, and stuff like that. Many of these complaints have EAI (today ESB) roots and are largerly irrelevant to SOA (including HTTP/REST). However, I see lack of contract as the main issue and perhaps even showst......

32 of 108 | Real SOA: Successes and Headaches - Sanjiva pointed to nice interview with Werner Vogels, CTO of Amazon.com. It's about real SOA in the large. Very good read. This is SOA: It has been a major learning experience, but we have now reached a point where it has become one of our main strategic advantages. We can now build very complex applications out of primitive services that are by themselves relatively simple. We can scale our operation independently, maintain unparalleled system availability, and introduce new services quickly without the need for massive reconfiguration. This is what ESB lovers should keep in mind: The first and foremost lesson is a meta-lesson: If applied, strict service orientation is an excellent techn......

33 of 108 | Real SOA: Successes and Headaches - Sanjiva pointed to nice interview with Werner Vogels, CTO of Amazon.com. It's about real SOA in the large. Very good read. This is SOA: It has been a major learning experience, but we have now reached a point where it has become one of our main strategic advantages. We can now build very complex applications out of primitive services that are by themselves relatively simple. We can scale our operation independently, maintain unparalleled system availability, and introduce new services quickly without the need for massive reconfiguration. This is what ESB lovers should keep in mind: The first and foremost lesson is a meta-lesson: If applied, strict service orientation is an excellent......

34 of 108 | Real SOA: Successes and Headaches - Sanjiva pointed to nice interview with Werner Vogels, CTO of Amazon.com. It's about real SOA in the large. Very good read. This is SOA: It has been a major learning experience, but we have now reached a point where it has become one of our main strategic advantages. We can now build very complex applications out of primitive services that are by themselves relatively simple. We can scale our operation independently, maintain unparalleled system availability, and introduce new services quickly without the need for massive reconfiguration. This is what ESB lovers should keep in mind: The first and foremost lesson is a meta-lesson: If applied, strict service orientation is an excellent......

35 of 108 | Real SOA: Successes and Headaches - Sanjiva pointed to nice interview with Werner Vogels, CTO of Amazon.com. It's about real SOA in the large. Very good read. This is SOA: It has been a major learning experience, but we have now reached a point where it has become one of our main strategic advantages. We can now build very complex applications out of primitive services that are by themselves relatively simple. We can scale our operation independently, maintain unparalleled system availability, and introduce new services quickly without the need for massive reconfiguration. This is what ESB lovers should keep in mind: The first and foremost lesson is a meta-lesson: If applied, strict service orientation is an excellent......

36 of 108 | Case Study: Starwood Hotels Uses SOA to Improve Guest Services and Cut Costs - Here is a case study on how SOA improves business and cuts cost. This success does not come with purchase of some products. It's rather about architectural vision and execution of it - how to identify candidates for services, how to build and how to share them. Simply, how to do SOA. Don't do nothing. Of course I'm glad that Systinet product plays important role in this particular case. But the key thing here is the change of IT folks mind. The EAI mind. Instead of endless data transformation among legacy apps - identify services, share/publish them, and use them. Any SOA product, including Systinet, is useless without this approach. ...

37 of 108 | Case Study: Starwood Hotels Uses SOA to Improve Guest Services and Cut Costs - Here is a case study on how SOA improves business and cuts cost. This success does not come with purchase of some products. It's rather about architectural vision and execution of it - how to identify candidates for services, how to build and how to share them. Simply, how to do SOA. Don't do nothing. Of course I'm glad that Systinet product plays important role in this particular case. But the key thing here is the change of IT folks mind. The EAI mind. Instead of endless data transformation among legacy apps - identify services, share/publish them, and use them. Any SOA product, including Systinet, is useless without this approach. ...

38 of 108 | SOA - Do Nothing? - This post (via Gervas) comments on Tim Bray's suggestion to ignore SOA. SOA is again put against REST. This is something I do not understand. So, is GData a service or not? Amazon? eBay? They all offer XML/HTTP APIs so I think Tim would agree they are 'Web-like things'. Yes, they are. But they are also services. It's WSA what Tim is talking about. Do you agree with this? (Enterprise architecture) EA = SOA + non-SOA (Service-oriented part of EA) SOA = WSA + WOA + XOA (Web Services Architecture) WSA = WSDL + SOAP + ESB (X-oriented architecture) XOA = somebody might miss corba, ejbs, (replacement for the X); not my cup of tea any more though (Enterprise Service Bus) ESB = No, tha......

39 of 108 | SOA - Do Nothing? - This post (via Gervas) comments on Tim Bray's suggestion to ignore SOA. SOA is again put against REST. This is something I do not understand. So, is GData a service or not? Amazon? eBay? They all offer XML/HTTP APIs so I think Tim would agree they are 'Web-like things'. Yes, they are. But they are also services. It's WSA what Tim is talking about. Do you agree with this? (Enterprise architecture) EA = SOA + non-SOA (Service-oriented part of EA) SOA = WSA + WOA + XOA (Web Services Architecture) WSA = WSDL + SOAP + ESB (X-oriented architecture) XOA = somebody might miss corba, ejbs, (replacement for the X); not my cup of tea any more though (Enterprise Service Bus) ESB = No, tha......

40 of 108 | WOA SOA - One of the many things I missed during my off-blog time is WOA. The acronym coined by Nick Gall of Gartner stands for Web Oriented Architecture. WOA emphasizes one particular architectural style, REST, as the best practice for a subset of SOA. Read some related posts here: by Dion Hinchcliffe or by Nick. Excellent. Web is a big laboratory where concepts are being proven by time and large-scale use. Then they go to enterprise. It was browsers and portals (intranets in general) in the past. And it's XML/HTTP nowadays - and it's usually called REST. I'm glad to read in these postings that what people build (whether by REST or by WSDL) is a service - and therefore part of SOA. Yes, there are......

41 of 108 | WOA SOA - One of the many things I missed during my off-blog time is WOA. The acronym coined by Nick Gall of Gartner stands for Web Oriented Architecture. WOA emphasizes one particular architectural style, REST, as the best practice for a subset of SOA. Read some related posts here: by Dion Hinchcliffe or by Nick. Excellent. Web is a big laboratory where concepts are being proven by time and large-scale use. Then they go to enterprise. It was browsers and portals (intranets in general) in the past. And it's XML/HTTP nowadays - and it's usually called REST. I'm glad to read in these postings that what people build (whether by REST or by WSDL) is a service - and therefore part of SOA. Yes, there are......

42 of 108 | Returning Back to Blogosphere - Hi everybody, I'm slowly returning back to the blogosphere. Systinet acquisition, introduction of new family of products for SOA Governance (Systinet 2), and kick off of new release cycles - these are good reasons to stay quiet for a while. I'm sorry I have thousands of posts in my rss reader and I'm unlikely to read them :-( I spent last week in Chicago - we had excellent sales meeting over there. I was in Chicago before but never had time to look around or visit the downtown. This time I did. Chicago takes the best from the coast cities as well as from 'the middle'. Wonderful city!...

43 of 108 | Returning Back to Blogosphere - Hi everybody, I'm slowly returning back to the blogosphere. Systinet acquisition, introduction of new family of products for SOA Governance (Systinet 2), and kick off of new release cycles - these are good reasons to stay quiet for a while. I'm sorry I have thousands of posts in my rss reader and I'm unlikely to read them :-( I spent last week in Chicago - we had excellent sales meeting over there. I was in Chicago before but never had time to look around or visit the downtown. This time I did. Chicago takes the best from the coast cities as well as from 'the middle'. Wonderful city!...

44 of 108 | Trust, SOA, and Salesforce.com - Cranky Salesforce.com users want updates, SLAs. I wrote many posts on SOA and Trust so I think no further comment is necessary....

45 of 108 | Trust, SOA, and Salesforce.com - Cranky Salesforce.com users want updates, SLAs. I wrote many posts on SOA and Trust so I think no further comment is necessary....

46 of 108 | Systinet Vision 2001 - In 2001, our investors pitch had five bullets: 1. Web services will be widely successful (database or application server-like) and will facilitate revolutionary change with an evolutionary step 2. We disagree with the hype around public UDDI. Web services will evolve ?privately? first. 3. Development of new applications will lead the adoption curve 4. Capturing developer mind share can be a crucial competitive advantage 5. Standards based versions of SOAP and UDDI will eventually commoditized, but being "first" with a strong implementation gives us temporary competitive advantage Roman gets sentimental from time to time ;-)...

47 of 108 | Formal Definitions - I haven't seen any formal definition of SOA and I don't think there is any. As there is no (afaik) formal definition of the term 'enterprise architecture'. At least formal in the sense Mark and Patrick are talking about. As far as 'formalism', I like Don's classic service orientation fundamentals. Others (e.g. Systinet) add some extra stuff like Registry as the architectural element of SOA. I like Registry-based SOA because I like Don's fundamentals and I believe Registry-based (or repository depending of your level of sophistication) SOA is perfectly aligned with them. But somebody else likes ESB, others are ok with EAI MOMs... There is no way to reach a formal agreement. Actually, I......

48 of 108 | Mercury To Acquire Systinet! - Read the press release here. I`m excited. Everybody I`ve met at Mercury is excellent and I`m looking forward to work on the new wave of SOA Governance....

49 of 108 | Bringing IT Challenges Down to Earth - "We chose Systinet`s registry technology for its strong standards-based support, including support of the UDDI V3 standard," said Burnett. "We also found Systinet Registry provided the flexibility we needed for our diverse user community." To read more on how NASA implements SOA go here....

50 of 108 | Case Study of Model Driven Engineering - Stefan Tilkov posted very nice http://www.innoq.com/blog/st/presentations/2005/SOA-MDE.pdf ">presentation on MDE including a case study. I would have two remarks: 1) I see little warning between two statements a) slide 20: A single, unified, enterprise data or process model would be great - but is next to impossible to achieve and b) slide 22: Vision: Create a unified SOA metamodel. Perhaps, the cost of creating unified metamodel (regardless whether SOA or whatever else) is higher than accepting the reality and work with multiple models directly without a metamodel? I understand creating unified metamodel should be (or rather it is) easier than unified model but... 2) There should be a ......