研究目的
To investigate the problem of data-intensive vNF service chain orchestration in inter-datacenter elastic optical networks, aiming to minimize the average service completion time.
研究成果
The proposed request sorting algorithm (SFEF) significantly reduces the average service completion time compared to benchmarks, and task rescheduling is beneficial in heavy background traffic scenarios. The algorithms effectively minimize buffering delays and improve resource utilization in inter-DC EONs for data-intensive vNF service chains.
研究不足
The study assumes a non-preemptive task scheduling model and one active instance per vNF per DC node, which may not reflect all real-world scenarios. The simulations are based on a specific topology (NSFNET) and traffic models, limiting generalizability. Task rescheduling is only effective in heavy traffic conditions due to insufficient spectrum fragments.
1：Experimental Design and Method Selection:
The study uses a sequential approach to solve the NP-hard problem, involving a request sorting algorithm (SFEF) and a data-intensive vNF-SC orchestration algorithm based on dynamic programming (DP) with and without task rescheduling. Simulations are conducted to evaluate performance.
2：Sample Selection and Data Sources:
Simulations use the 14-node NSFNET topology as an inter-DC EON, with dynamic background traffic and pre-deployed vNFs. Data-intensive vNF-SC requests are generated using a Poisson process over 2000 time slots.
3：List of Experimental Equipment and Materials:
The setup includes bandwidth-variable optical switches (BV-OXCs), bandwidth-variable transponders (BV-Ts), and local data centers. Lightpaths are established with up to two per DC pair, each occupying 11 frequency slots.
4：Experimental Procedures and Operational Workflow:
The algorithms are implemented in Matlab R2013a. For each request, the algorithms select DC nodes for vNFs, assign lightpaths, and schedule data transfers and processing tasks. Performance is measured in terms of average service completion time under light and heavy background traffic scenarios.
5：3a. For each request, the algorithms select DC nodes for vNFs, assign lightpaths, and schedule data transfers and processing tasks. Performance is measured in terms of average service completion time under light and heavy background traffic scenarios.
Data Analysis Methods:
5. Data Analysis Methods: Results are analyzed by comparing average SCT across different algorithms (SFEF vs. SLVF, with and without task rescheduling), and distributions of delays (data processing, bulk-data transfer, buffering delays) are examined.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
