the reliability of the middleware is influenced by what factors? How to review the reliability?
CodePudding user response:
Didn't specifically done, said his own understanding, welcome to discuss and correct, and feel do much of the background and foreground development would eventually involve some, with the average development is also very obvious, the difference between a common development is geared to the needs of the business, what functions to implement one by one, like the tree to open branch leaf, common development to realize every each branch leaves, middleware for the trunk, transport nutrients, and the leaves don't need to know where to nutrient, only know that this nutrient is what he needs, what route nodes which passed by the middleware in what way to calculate the analysis and choice of all the pages or branches to transport through the unification of the corresponding interface, access (unified interfaces and protocols to ensure that the adaptability of different platform),As to guarantee the middleware in the adaptability of different operating system platforms and the balance of performance, it depends on different according to different platform middleware call apis and the relevant technology of the corresponding platform, take the thread, the file system, a network interface for different platforms have different platform API this process involves using the development of open source or its components, the complex operation, cross-platform, agreement of what are encapsulated, provides a simple interface calls, call these interface to realize the function of the relevant business development, without considering the specific implementation details, also need not consider operating system update, version updates,
Earlier this work like some company has its own underlying library, the function interface are encapsulated, convenient later developers use direct call, now these underlying library still exist, but a lot of company scale, business came up, distributed multi-platform deployment, simple encapsulates the underlying library can't satisfy the requirements, need to encapsulate some IO intensive, with a dense network operations, or the operating system, became the biggest bottleneck need to develop some protocol stack or bypass operation system kernel component based development, caching and network middleware development will achieve cross-platform, high concurrency, actually middleware was born very early, but now is no longer only in a few company's patent, each company can according to their own business needs to develop their own middleware,
Reliability affected by what I can think of is to scale and hardware platform, the scale of the hundreds and thousands of servers and the concurrency value, involving the technology need to be solved and the bottleneck of exposure and is different also, I want to big company product reliability are detected in the wars again and again, and the theoretical framework and solve the problem of war when suddenly the situation again and again honed, according to different application scenarios to highlight different skills