动态重构_05

《Software-Defined Efficient Service Reconstruction in Fog Using Content Awareness and Weighted Graph》
《利用内容感知和加权图在雾中软件定义高效的服务重构》

问题归结：
it’s hard to enhance resource-efficiency in the existing relatively static and proprietary framework of fog nodes due to the diversity of service requirements. With the growing deployment of fog computing, the overall resource consumption in fog will be huge without considering the efficient service provisioning in each fog node. On one hand, different fog users require diverse local services policies which are carried out within the fog nodes.
由于服务需求的多样性，很难在现有的相对静态和专有的雾节点框架中提高资源效率。随着雾计算的部署不断增长，如果不考虑每个雾节点中有效的服务供应，雾中的整体资源消耗将是巨大的。

结果方案：
We propose a software-defined efficient service reconstruction (SDSR) scheme in fog using content awareness and weighted graph. Service reconstruction mechanism is devised to dynamically recycle modularized resources after mapping different contents to relevant operations. Weighted graph is introduced to schedule and optimize the services reconstruction in terms of resource saving during content-driven controlling.
我们提出了一种使用内容感知和加权图的雾中软件定义的高效服务重建（SDSR）方案。服务重构机制被设计为在将不同的内容映射到相关操作之后动态回收模块化资源。引入了加权图，以在内容驱动控制期间节省资源方面计划和优化服务重构。

雾计算：
Massive data collected by the large-scale end devices (i.e. smart sensors, road side units and radio frequency identification systems) are directly sent to the local fog nodes within their region to get timely and high-rate services instead of transmitted to the remote cloud. Finally, the processed contents are uploaded to the upper layer fog users for further global analytics as well as long-term storage.
大型终端设备（例如，智能传感器，路边单元和射频识别系统）收集的海量数据直接发送到其区域内的本地雾节点，以获取及时，高速率的服务，而不是传输到远程云。最后，将处理后的内容上传到上层雾用户，以进行进一步的全局分析以及长期存储。

为何回收资源：
Although the services required by different fog users and contents are diverse, they have many overlapping parts in component operations. This characteristic offers a promising way to recycle these basic units.
尽管不同的雾用户和内容所需的服务各不相同，但它们在组件操作中有许多重叠的部分。此特性提供了回收这些基本单元的有希望的方法。

优势：
Fog users are enabled to program and customize services performing in the underlying fog nodes by reconfiguring particular demands or modifying service modules via the exposed interfaces instead of creating new services to adapt to their changing requirements.
雾用户可以通过重新配置特定需求或通过暴露的接口修改服务模块来编程和定制在底层雾节点中执行的服务，而无需创建新服务来适应他们不断变化的需求。

方法简介：
The modularization of basic service operation is important for the recycling and reusability of resources. Services are decomposed into multiple basic essential operations. An atomic operation (such as data compression, data filtering, data averaging computing, etc.) is further abstracted and encapsulated into modules, each of which is implemented by different algorithms. All the service modules in the charge of Module Manager are stored in the service module pool. They can be reused and easily migrated when the application scenario transforms or fog users’ requirements change.
基本服务操作的模块化对于资源的回收和可重用性很重要。服务被分解为多个基本的基本操作。原子操作（例如数据压缩，数据过滤，数据平均计算等）被进一步抽象并封装到模块中，每个模块由不同的算法实现。由模块管理器负责的所有服务模块都存储在服务模块池中。当应用程序场景发生变化或模糊用户的需求变化时，它们可以重复使用并轻松迁移。
The service reconstruction involves invoking and initializing the required modules and scheduling them in the reasonable
order. These recyclable service modules are stored in the service module pool in the Service Resource Repository, while the execution logic order is maintained in the Content-based Policy Library. These are done by the Service Reconstructor. Since modules are designed loosely coupled, the Service Reconstructor is introduced as a middleware to allow multiple independent modules to be “plugged in” to it [14].
服务重建涉及调用和初始化所需的模块，并以合理的顺序对其进行调度。这些可回收​​服务模块存储在服务资源库中的服务模块池中，而执行逻辑顺序则在基于内容的策略库中维护。这些由服务重构器完成。由于模块的设计是松散耦合的，因此将服务重构器作为中间件引入，以允许将多个独立模块“插入”到其中[14]。

调度问题：
In most cases, we tend to choose the policies with the least resource consumed for the service provisioning. But sometimes users may have special requirements. For instance, they may demand one of the component operations performing on certain content implemented with a specific module(algorithm). This customization can be realized by setting preferences on relevant items via the interfaces provided by user-defined reconfiguration layer in the label-operation mapping table stored in the Content-based Policy Library to enforce the execution of the required service modules. Under these circumstances, the Requirements Analyzer checks the constrains placed on the certain labels to guarantee the needed modules involved to achieve the user-defined service reconstruction.
在大多数情况下，我们倾向于选择用于服务供应的资源最少的策略。但是有时用户可能有特殊要求。例如，他们可能要求对使用特定模块（算法）实现的某些内容执行的组件操作之一。可以通过基于用户的重新配置层在存储在基于内容的策略库中的标签操作映射表中由用户定义的重新配置层提供的接口来设置首选项，以实现所需服务模块的执行，从而实现此自定义。在这种情况下，需求分析器会检查放置在某些标签上的约束，以确保涉及实现用户定义的服务重构所需的模块。

个人总结：
不但要考虑周围环境以及用户需求的变更，也可以考虑系统分配时的资源调度问题。通过加权图对各个服务的资源调度量进行可视化。
论文表明，对于服务的重构，我们可以将单个服务进行划分，划分为子服务。利用不同的实现方式，对各个子服务的算法进行实现。可以在子服务层次上对不同的子服务进行重构，来重新构成新的服务。