FIB表急剧增长是互联网高速发展面临的重要问题之一,FIB表分解存储能有效解决该问题.现有的SPAL技术将FIB表较均匀地分解存储在不同线卡,但仍然存在较多的表项冗余存储现象,并且实现复杂.对此设计了一种新型的转发表分解存储模型(Decomp...FIB表急剧增长是互联网高速发展面临的重要问题之一,FIB表分解存储能有效解决该问题.现有的SPAL技术将FIB表较均匀地分解存储在不同线卡,但仍然存在较多的表项冗余存储现象,并且实现复杂.对此设计了一种新型的转发表分解存储模型(Decomposed Storage of FIB,DSF),它依据IP前缀的前若干bit位实现线卡对转发表的分解存储,并只带来极少的冗余存储.DSF的改进方案———EDSF,更可使各线卡非常均衡地完成分解存储.提出的分解存储模型缓解了FIB表项急剧增长问题的解决压力,同时大大节省了硬件资源.对于线卡数量更多的可扩展路由器尤为适合.通过对当前运营的路由表的分解存储实验研究及与其它方案的比较,验证了文中模型良好的存储性能.展开更多
路由表的急剧扩张使得可扩展分布式路由器中线卡的硬件成本迅速增长。提出一种路由表的均衡存储转发模型RESD(Storage and Decomposition Model based on Even Routing),它属于非完全存储方式。RESD模型通过初始化分配将长度相邻的前缀...路由表的急剧扩张使得可扩展分布式路由器中线卡的硬件成本迅速增长。提出一种路由表的均衡存储转发模型RESD(Storage and Decomposition Model based on Even Routing),它属于非完全存储方式。RESD模型通过初始化分配将长度相邻的前缀尽可能地分配到相同或是相邻的线卡上,然后通过分段式分配均衡每个线卡上的路由数目,最后通过路由特征位信息表和定位路由来实现路由信息的查找。实验证明,该模型可以在路由表分解存储均衡的同时大大减少LPM匹配中各线卡的并行查找次数。RESD模型的设计和实施原型都是以真实的环境为基础的,因此,RESD模型与现有网络的体系结构具有良好的兼容性,易于实施。展开更多
In recent years, due to the widespread use of electronic services and the use of social network as well, large volumes of information are being made that this information contains various types of things such as video...In recent years, due to the widespread use of electronic services and the use of social network as well, large volumes of information are being made that this information contains various types of things such as videos, photos, texts etc. besides large volume. Due to the high volume and the lack of specificity of this information, covering them through traditional and relational databases is not possible and modem solutions should be used for processing them, so that processing speed is also covered. Data storage for processing and the way of accessing to them in memory, network communication, covering required features for distributed system in solutions that are in use for storing big data, are the items that should be covered. In this paper, a collection of advantages and challenges of big data, special features and characteristics of them has been provided and with the introduction of technologies in use, storage methods are studied and research opportunities to continue the way will be introduced.展开更多
文摘FIB表急剧增长是互联网高速发展面临的重要问题之一,FIB表分解存储能有效解决该问题.现有的SPAL技术将FIB表较均匀地分解存储在不同线卡,但仍然存在较多的表项冗余存储现象,并且实现复杂.对此设计了一种新型的转发表分解存储模型(Decomposed Storage of FIB,DSF),它依据IP前缀的前若干bit位实现线卡对转发表的分解存储,并只带来极少的冗余存储.DSF的改进方案———EDSF,更可使各线卡非常均衡地完成分解存储.提出的分解存储模型缓解了FIB表项急剧增长问题的解决压力,同时大大节省了硬件资源.对于线卡数量更多的可扩展路由器尤为适合.通过对当前运营的路由表的分解存储实验研究及与其它方案的比较,验证了文中模型良好的存储性能.
文摘路由表的急剧扩张使得可扩展分布式路由器中线卡的硬件成本迅速增长。提出一种路由表的均衡存储转发模型RESD(Storage and Decomposition Model based on Even Routing),它属于非完全存储方式。RESD模型通过初始化分配将长度相邻的前缀尽可能地分配到相同或是相邻的线卡上,然后通过分段式分配均衡每个线卡上的路由数目,最后通过路由特征位信息表和定位路由来实现路由信息的查找。实验证明,该模型可以在路由表分解存储均衡的同时大大减少LPM匹配中各线卡的并行查找次数。RESD模型的设计和实施原型都是以真实的环境为基础的,因此,RESD模型与现有网络的体系结构具有良好的兼容性,易于实施。
文摘In recent years, due to the widespread use of electronic services and the use of social network as well, large volumes of information are being made that this information contains various types of things such as videos, photos, texts etc. besides large volume. Due to the high volume and the lack of specificity of this information, covering them through traditional and relational databases is not possible and modem solutions should be used for processing them, so that processing speed is also covered. Data storage for processing and the way of accessing to them in memory, network communication, covering required features for distributed system in solutions that are in use for storing big data, are the items that should be covered. In this paper, a collection of advantages and challenges of big data, special features and characteristics of them has been provided and with the introduction of technologies in use, storage methods are studied and research opportunities to continue the way will be introduced.
