Recently, due to project requirements, some middleware, such as ice, Ace, and Globus, need to be checked. Ice is a set of alternatives developed after a few days, out of being too tedious and tired of CORBA. It should be pointed out that, CORBA is a standard developed by OMG, a member organization of hundreds of enterprises. It is relatively bureaucratic and has many implementations. Its Commercial Products include Iona, Mico, and ace tao for free, many people have been discouraged by the complicated CORBA. I remember that I had read this section "C ++ advanced programming" twice or three times before I knew it, I do not know that the author of this book is one of the designers of ice. it is conceivable that ice was born from its birth to replace CORBA and has been widely recognized in the past few years. It seems that, once again, this proves that the technology advances constantly from a few days, and the simplicity is the beauty. This relationship is somewhat like EJB and spring. Ace is a set of underlying communication encapsulation mechanisms. It can be used to develop network communication programs. Ace should be a concept better than ice, the former can be a better SDK than the socket, and the latter is a distributed object model. It is much easier to develop a network application, because Ace is too perfect and the learning curve is steep. Globus is a standard implementation middleware for grid. Grid has been studied for a long time before, and the technical community tends to do this. Once a new concept is introduced, it will be easy for the technical community to upgrade in some ways, the grid is also used to create one bubble after another. I opened the "Computer World" newspaper five years ago. There were also several special editions that introduced the grid and thought it was the third Internet wave after the Web. Now it seems that this has become a joke, web 2.0, which satisfies user needs, has become a hot topic-essentially, it is still a bubble, but the difference with the grid is that Web 2.0 is user-centered, the concept of increasing user experience is summarized from new applications that are constantly emerging on the Internet. Even if this bubble breaks down in the future, this trend of improving user experience cannot be reversed; the grid is different. Similar to CORBA, this is also a concept developed by academia (which is an industry). Its vision is good, but it is too advanced in actual use. When studying the source code of globus 2.0 a few years ago, we found that this is essentially an extension of previous Cluster Computing. Fortunately, over the past two years, with the entry of IBM and other enterprises, GRID has also begun to develop in a pragmatic direction, more positioned in enterprise-level services, and gradually integrated with SOA. It should be noted that the latest version of ice 3.0 has a grid extension, but this extension is only used to better expand the discovery and scheduling of distributed computing resources, so as to better provide enterprise computing capabilities.
I want to investigate these things because my platform has encountered some problems and needs to be expanded. The Platform runs as follows: the server provides Web-based computing services. However, due to heavy computing tasks and non-reentrant computing threads, I have to use synchronization to ensure security services. In the future, as the number of machines increases, the existing architecture can meet the needs of simple expansion in the short term, but does not have good load balancing and dynamic expansion capabilities. Therefore, it needs to be improved. After preliminary research, I decided to adopt the ice architecture, a good learning curve, a platform language independence and a concise design were the key to my decision. I will write some relevant details here when the building is complete.
