Hello Ron I was wondering if you could provide some references to the current work done on ISB as mentioned in this two webcasts, in particular architectural diagrams, etc. I have done quite a bit of work on this area and would like to compare notes sort of speak. It also would be good if we get some contact informatin and ways to participate in the ISB project Thanks Ron and keep u p the good work Luis Gonzalez
Hello Ron, Interesting show to say the least, it basically shows how inmature the concept of service bus and how experimental SOA applicability still is even today. By the way one of the oldest IEEE standards in electronics still in use today is the IEEE 488 standard also known as HPIB (Hewlet packard interface bus) or GPIB (General purpose Interface bus). This is a slow speed standard that allows test instruments of different brands to perform a test on a device (patient), test instruments are controled by a controller that handles the bus, and the controller communicates with a higher level language progam that actually describes the test to be performed. This standard was invented by HP many years ago, but we can draw from it many concepts about diversity of participating components, orchestration, and reasons of having a bus and a controller. Every single test instrument could be considered a web service that could serve in many different bus configurations (test), etc. I could go for ever on this topic but I just wanted to mention it to you because this standard has lasted for so long, and my guess that it is simply because it works, It has a clearly defined message handshake, interface messages, device messages, and local messages. Okay I will stop here for now Here are some comments drawn from my own experience on SOA. I have created coarse granular SOA in the pass for a travel agency: - availability services: one service per segment ((one for hotels, one for air, one for cars), - reservation services: onse service for all the segments: it handles booking, change, and cancel. Some components like the adapters to third party providers utilized during availability are also utilized here but as reutilized components and not as separate services. - Security services: handles what we call a user set, a user set can have many users in it: employees, business partners, and/or customers. One user in the user set is the principal user, a user set also has at least a target user that can be a target customer, target employee, or target partner, or a combination. A user set is very coarse granular since it contains multiple users, and a user in itself has many components: and identity component, an authorization component and a profile component. It works pretty good actually and it is very flexyble. - Data mapping services: it is a single service capable of providing dtat maping info for any adapter and also handles errors asynchronously, and data maping administration. - Credit Card Services: is a wraper service around a legacy credit card system and provides credit card functionality to any segment (hotels, air, car, etc.) - Message Logging services: allow for the centralized storage of error messages regardless of where the error physically happen - Notify now services: this one is capable of preteing up messages and sending the messages through different channels as required. - Work flow services: a service that behaves more like a visitor than an imposition in other services. - Content services - Master reference services - Pricing services - Etc. All these services are coarse in nature and they handle a set of data composed of multiple collections (tables) in other words their interface is not chaty at al, one set of data contains enough information for the consumer to have at it. and each service may handle large distributed transactions. Cordially, Luis G Gonzalez Architect Las Vegas
I downloaded the 774,488 KB wmv for the Pirmaid Patterns but all I got was your introduciton at the beginning and your comments at the end, all in between was just a little bit of music. I try the file both with Nero and Windows Media Player
Luis G Gonzalez
Las Vegas, NV
Hello Ron, The CommSee arcast were very interesting and I hope that there will be white papers and more inofrmation about the system and their architecture. I have many many questions at all levels about the CommSee project but would like to read more about it before asking them. Are or would there be any sources available that you know of? So for now I wanted to ask you a simple question. I kind of heard of EDRA before, but when I went into looking for recent information about EDRA I found practically nothing. Is the Services Factory replacing EDRA or is it an improved EDRA, are these two separate projects evolving today? A very genral question is what would CommSee do different with the new technology of today? inparticular Communication Foundation, and WorkFlow foundation? Also I still favor strongly typed internal entites instead of dealing with pure xml across all layers. It seems that CommSee is dealing with XML all the time. I believe the benefits of strongly typed representations may have a cost on serialization and deserialization but probably a typed object is much smaller than an xml object, also type manipulation is more performat, simplifies maintenance and error detection at compilation time. It s doubful tha txPath and is text search engine could be as fast as traversing a stronly type collection. etc etc. I believe we agree but your comments will help clearing the issue. Hopefully one of this days there will be a mano a mano between utilizing strongly typed dataset and other type of objects as boundary entities. Also at some point I thought that datasets were a w3c standard as well. There are too many advantages about utilizing typed datasets over custom collections and maybe an arcast where you play evils advocate against your own optinion could help. Thank you Ron and keep up the good work Cordially, Luis G Gonzalez Architect Las Vegas
That was a very good Arcast / web cast with Roger Wolter.
I bought his book as well, it seems to me that with Service Broker some middle-ware layers could be reduced or moved closer to the database, improving performance, queuing reliability, etc.
I found interesting that his book does not mention architects at all, just DBAs and DB developers, and it seems to me that this could become an important solution to implement many architectural patterns. I was wondering why didn't he mention architects, and also the scope of the applicability of the solution seems humbled to me, meaning that I believe it has many more applications than just what is suggested, I heard many times that it only apply to Database applications, and I do not know exactly what does that mean. Maybe I am totally wrong, so I was wondering about your take on this, how does it relates to architecture directly and its applicability in normal solutions, not just database applications, its applicability in implementing several enterprise architectural patterns.
On a different now, In this new Arcast web page, I also had to create a profile in order to write this message, and curiously enough the profile options only mention Solutions architect and Infrastructure architect, not Enterprise architect… ???
When it comes to topics of Arcast, some times it will be interesting to look deep into the applicability of some patterns, like:
- How to create application Context that is shared in a distributed environment.
- How to utilize work flow as a visitor pattern in a distributed environment.
- Much more on applicability of strongly typed data sets, Opinions, etc. How compatible are strongly typed datasets between 2003 and 2004, etc. I believe you are not in favor of strongly typed dataset, which are a w3c standard, but honestly the more I utilize them the more I get from them. SOA tenants and SO tenants differ a little according to some, and I believe in some instances strongly typed datasets are an excellent messaging alternative. I am interested on positive opinions about strongly typed datasets, if you can find some for me that will be great.
- All those cool patterns that the P&P group is utilizing, like going deep into CAPs or the Object generator, maybe I am not utilizing the right terms or spelling but you know what I am refereeing too.
- Applicability and adoption of Attributes, difficulty of adoption and understanding of new concepts like generics in c#, etc.
- It is clear that Microsoft aims at a large base of developer types, and architecture probably takes less advantage of quick and no so clean solutions. Identifying those are also another interesting topic. In particular the case of declarative coding, strongly typed coding as oppose to quote delimited coding. I am very much against quote delimited coding because of how expensive maintenance of this is. I like having automated tools that generate types out of quote delimited data, this way any time you change the quotes, the types change as well, or fail if you don’t change them.
I believe the work you are doing is very valuable to all of us, so thank you very much to you and Microsoft
I have listening and studying most of the things you have been involved with for the last 4 years. Starting from the blocks, webcasts, etc/ and now the arccasts.
It seems to me that your presentation skills have improved quite a bit so I wanted to congratulate you for that. Honestly I did not like all that fake and fun stuff that you use to do a little bit too much before, now it seems to be a good mix, I agree with you that things should not be taken too seriously because then we forget about how much fun it is to do it. But the content and the information you provide have been very valuable to the companies I have worked for and myself, so thank you very much.
The reason of the present is to ask you for contact information for Danny Boyton (name speeling as well), the person you interviewed in this arccast. It seems that him and I have taken very similar paths when it comes to architecture.
Since before PDC in LA I have been wondering about workflow and Coarse Interfaces common to xml web services. I see work flow as a visitor pattern to many applications probably provided as an xml web service in itself. It seems to me that workflow foundation works better with finer granular interfaces in particular when you want to handle transactions and all that. I did not get a chance to implement the workflow visitor pattern concept I had in mind, but It is still something I want to figure out completely. Id did get to develop a workflow as a service utilizing strongly typed datasets as the state machine, it works very well and it is easy to visualize. The dataset not only provides the state but also a dataset derived class provides the engine, that can be consumed by any lager.
So there it is, workflow as a visitor is mostly and important visitor that captures and provides information that help the business logic control its own transactions (because of coarse granular interfaces, the buisnes logic does quite a bit). Yes the workflow controls long-lived transactions whose roll back procedures are not left to the underlying structures (ado.net, .net framework, sql server, etc.) but to the workflow itself and to additional business logic.
Anyhow I would appreciate very much if you provide me with Danny Boyton's contact information. I am not even sure how to spell his name; after all it is an arccast. right?
Thanks Ron and keep up the good work.
Luis G Gonzalez
self proclaimed enterprise architect, systems architect, and applications architect.