Slashdot Mirror
Survey: SOA Prominent On 2005 budgets
Michael S. Mimoso writes "A Yankee Group survey of 473 enterprise decision makers reveals that companies have put aside money for service-oriented architectures for 2005." This is a bigger deal than it sounds - if companies keep moving this away, it will mean a sea change in corporate technology usage - and change the way/why development is done. We're talking everything from SOAP stuff (ITMJ is part of OSTG) to wholesale ASP adoption like Salesforce.com.
Its Service-Oriented Architecture
A quick Google search will give you a reasonable answer.
SOA is Service-Oriented Architecture. Makes sense doesn't it?
Lots of comments on the buzzwordiness of SOA, and questioning the technical merit. I've been working on a SOA project for a couple months now, and I can tell you - the technical merit is there (as well as the acrid stench of buzz).
The core idea of SOA is that there are a lot of enterprises out there with lots of legacy databases on their networks. They also have small, decentralized app development teams that just want to put the data in front of the customer, as quickly as possible. Allowing all those teams direct access to all those databases is both expensive and risky (from a security standpoint) and expensive and difficult (from the front end developer's standpoint). SOA is a way to put a single point of entry across multiple databases. The front end people can code hellbent for leather against SOAP, without thinking about security or SQL, while the SOA team writes at a somewhat slower more methodic pace, linking in security (perhaps via LDAP) and handling handling the SQL.
Basically it's a way of keeping the O/R mapping and database security problem with a single team, while also allowing individual departments and divisions of the corporation to have their own app development teams.
Stop-Prism.org: Opt Out of Surveillance
Chicken Little here with an announcement... Some feel that Microsoft has actually lost the war for the Windows API
/.
Here is the train that no one sees coming - MS isn't stealing an API or a package or a language this time. With Indigo, MS is trying to steal the next major programming paradigm. MS is quietly patenting the successor to OO, every wild and not-so-wild idea that anyone can pull out of their ass is being patented as a means of covering all bets. Somebody who works for MS (one of the people who burdened us with XML and then came over from the XML dark side, only to take the SOA ring back with him to the MS dark side) realized that OO programming is too hard and inefficient (save your sob stories) and said wow this old school internet thing was right all along, you remember they used have stuff like UUCP, SMTP, etc. So they slap a new name on it and all the middleware vendors dogpile onto it - it is now called SOA - A nice quippy 5 point definition for you here:
1) Interactions between "services" is by passing messages (not fucking objects) using a "contract" [protocol was too computer-y sounding] OMG, sounds almost platform independent, but wait don't forget the patentable XML trick - that will un-platform independent it.
1b) Ability of interacting services to hold a meaningful conversation is based on protocol version compatibility. Oops, I mean it is based on "Policy" (Don Box) This is also known as "Duh!"
1c) Services are autonomous. This is also known as "SOA Fantasy: The Duhpocalypse"
2) Assume an unreliable network - that was easy back then.
3) Look shit ("services") up in some kind of directory (DNS was too flexible and easy, lets go with UDDI or something similarly unimplementable or Active Directory maybe?)
4) Have boundaries (used to be called "You have your systems and we have ours", aided by a thing called dns to define the "boundaries").
5) Try to be stateless - good luck.
The only problem for people like BEA who sees that they will lose their business if people go to a fully distributed SOA model for apps in the future is: MS has gotten religion in time, this time, and BEA can't compete with MS's devtools and their ownership of a ubiquitous server platform (ie every system running Indigo - which is included "free" with any MS OS) and it fits perfectly with their goals:
1) Make it easy for any idiot to develop and deploy MS apps
2) Force people to run clients and servers using only MS OSs
3) Disallow FOSS software from duplicating functionality or participating as clients or servers
I think that covers their goals.
And when they patent everything to do with SOA and roll out Indigo they will have achieved that.
I don't think there is anything that will keep them from this. Anyone? anyone?
Copyright 2004 Rob iQEVAgUBQTBHL4GVnbVwth
Attribution required for reproduction outside of
Why the "uncool" copyright? because every techno-hipster is blathering on and on about SOA these days and I don't want to see my post show up on their "advertise here for $100/wk" techno-histper blogs without proper attribution, even if they do only want to make fun of it.
Dynamic service registration
Dynamic service discovery
Support for one or more standard protocols for service invocation
Note the absence of the acronyms "SOAP" and "XML" on that list.
Patrick