Just as early erp soa was pushed to the "God" platform again

Just as early erp soa was pushed to the "God" platform again
 
Source: zhidianwang Author: Shi Jiang

There are still many unsolved questions about the service-oriented architecture. Many people, even in the IT field, say they do not understand what SOA can do or how to deploy SOA. SOA has been exaggerated to an incredible level by software vendors and analysts, but there is little valuable information about the basic meaning of this new method.

How can soa fail when no one really owns SOA?

The following lists some unsolved questions about SoA:

1. If SOA fails, where are these "horrible" stories?

Many scholars and analysts claim that SOA will be a failed idea; no one has completed the deployment of SOA yet. At one time, everyone could not wait to announce that SOA was half-dead. However, what I saw or my own investigation showed that, most companies are still planning or considering their first service-oriented project. As a matter of fact, the major challenges that I have heard about SoA over the past few days are: SOA is too successful, and in those enterprises that are trying to deploy SOA, too many services are being added or created --- or required. This is why so many vendors have publicized SOA governance.

2. How do people know the time when the SOA project is successful or fails?

The paradox between SOA is that those enterprises that are most inclined to adopt SOA are exactly the enterprises with the lowest SOA requirements. The minimum value that is needed. If these enterprises are far-sighted in management, they may also deploy many other applications in the enterprise system to provide support for SOA, such as business intelligence and analysis, data warehouse, and customer relationship management. How much of their ongoing success can be attributed directly to SOA? What is the definition of success? Cost saving? What is a single end-to-end process provided by Web Services?

This is one of the first major challenges facing SOA-success is a long-term process. The success of SOA lies in the ability of enterprises to share services across multiple business units when the service development time is significantly reduced; or, due to the increasing flexibility of the enterprise's underlying infrastructure, this greatly reduces the time for enterprises to reconfigure products or services and bring them to the market.

However, the only criterion in the market that truly measures long-term success is the increasing revenue and stock value of enterprises. In addition to SOA, there are many other factors that may affect this. The real problem lies in figuring out how to measure the contribution of SOA to enterprise success. The "success" of SOA itself is unrelated to this.

3. How many full-featured real SOA deployment?

According to some analysts, many (75% or more) companies are currently implementing SOA projects. Some other analysts said that only 4% of enterprises deploying SOA are currently deployed. What are their standards? Depends on the number of services or the granularity of these services? How many applications or processes can access loosely coupled components with service functions? Is it just a bunch of Web services that have become the State Oceanic Administration? Web Services may need better "care"-governance, registration, management, and so on, so they will become more SOA-oriented. But what is this boundary?

4. How software vendors instill in users a concept that makes it easy for users to use their products.

Is it a blessing or a curse for software vendors? The real benefit of SOA is that services can be exchanged as needed. Users may use the SOA components of this vendor today, and may use the SOA components of other vendors tomorrow. This is also one of the challenges that software vendors are facing, especially those who strongly advocate SOA. (Of course, all vendors now say they are all related to SOA, right ?)

5. Who pays for SOA?

Which department will spend a lot of money and manpower to build such a system that will be used by anyone else? Other departments can use the services provided by the system without spending any resources. Applications with SOA functions may require more costs than traditional point-to-point interfaces in the initial stage, and ROI will be reflected in economies of scale. As Ma markrix explained, in the long run, the economies of scale produced by SOA can bring a better return on investment compared to point-to-point applications with lower cost for early implementation. However, risks generally occur when enterprises think they have deployed SOA, but in the end there is no ROI or very low, because what they deploy is not a real SOA-it is still a point-to-point interface. Who will take these risks into account? Or who is asked to consider these risks?