The rapid growth of deep learning has created unprecedented demand for computing resources, while many small and enterprise-level clusters remain underutilized. Computing resource exchange platforms offer a solution by aggregating these idle resources. However, effective cluster-task matching depends on accurate performance prediction. Existing approaches, which decouple prediction from matching, often lead to suboptimal decisions due to misaligned objectives. We propose a Matching-Focused Cluster Performance Predictor (MFCP), an end-to-end framework that integrates performance prediction with task matching to improve decision accuracy and resource utilization. Unlike existing methods that prioritize prediction accuracy, MFCP minimizes decision regret by aligning the predictor’s loss with optimal matching objectives. To handle non-differentiable matching optimization, we use continuous relaxation and incorporate constraints via an interior-point method, ensuring meaningful gradients for training. For non-convex optimization, we approximate optimal decisions with gradient descent and estimate gradients using zeroth-order perturbation. Experiments show that MFCP consistently outperforms existing methods across different cluster environments and scales, achieving lower matching regret and higher resource utilization.

IPv6 has shown notable growth in recent years, imposing the need for high-speed IPv6 lookup. As the forwarding rate of virtual switches continues increasing, software-based IPv6 lookup without using special hardware such as TCAM, GPU, and FPGA is of academic interest and industrial importance. Existing studies achieve fast software IPv4 lookup by reducing the operation number, as well as reducing the memory footprint to benefit from CPU cache. However, in the situation of 128-bit IPv6 addresses, it is challenging to keep both operation numbers and memory footprints small. To address the issue, we propose the Neurotrie data structure, which supports fast lookup and arbitrary strides. Thus, a good balance can be made between trie depth and memory footprint by computing the proper stride for each Neurotrie node. We model the optimal Neurotrie problem which minimizes the depth with limited memory footprint and develop a pseudo-polynomial time baseline algorithm to construct Neurotrie using dynamic programming. To improve the performance and reduce the computation complexity, we develop a deep reinforcement learning-based approach, which leverages a deep neural network to construct Neurotrie efficiently, based on characteristics captured from real IPv6 prefixes. We further refine the data structure called Neurotrie-S and develop an efficient mechanism for routing updates. Experiments on real routing tables show that Neurotrie-S achieves a lookup rate 34% higher than that of state-of-the-art approaches. We implement a Neurotrie-based software switch, and the forwarding rate of Neurotrie-S is about 10% to 345% higher than other algorithms.

IPv6 has shown notable growth in recent years, imposing the need for high-speed IPv6 lookup. As the forwarding rate of virtual switches continues increasing, software-based IPv6 lookup without using special hardware such as TCAM, GPU, and FPGA is of academic interest and industrial importance. Existing studies achieve fast software IPv4 lookup by reducing the operation number, as well as reducing the memory footprint so as to benefit from CPU cache. However, in the situation of 128-bit IPv6 addresses, it is challenging to keep both operation numbers and memory footprints small. To address the issue, we propose the Neurotrie data structure, which supports fast lookup and arbitrary strides. Thus, a good balance can be made between trie depth and memory footprint by computing the proper stride for each Neurotrie node. We model the optimal Neurotrie problem which minimizes the depth with limited memory footprint and develop a pseudo-polynomial time baseline algorithm to construct Neurotrie using dynamic programming. To improve the performance and reduce the computation complexity, we develop a deep reinforcement learning-based approach, which leverages a deep neural network to construct Neurotrie efficiently, based on characteristics captured from real IPv6 prefixes. We further refine the data structure and develop an efficient mechanism for routing updates. Experiments on real routing tables show that Neurotrie achieves a lookup rate 34% higher than that of state-of-the-art approaches.

Fast rerouting (FRR) has become an important technology for networks to enhance the resilience against failures. On the other hand, there is an increasing interest on integrating routing control and network functions, enabling highly programmable, adaptive, and cross-layer networking. In this paper, we propose a unified FRR scheme (UFRR) under the circumstance of network function virtualization and service function chain (SFC). By exposing the SFC information to the network layer, UFRR can make rerouting decisions that are more flexible, and protect the routing against various types of failures, including switch, router, and server failures. We propose the system architecture, and develop two algorithms to compute rerouting paths, which consider path length and forwarding entry number concurrently. Simulation results show that the unified FRR is effective, and the path stretch can be reduced by more than 10% compared to a typical network layer FRR scheme, while the forwarding entry number remains similar.

Network function virtualization has become a promising technology recently. To accommodate dynamic traffic such as burst, NFV can scale virtual network functions (VNFs) across several physical servers, which is known as horizontal scaling, or enlarge the capacity of a VNF leveraging idle resources in one physical server, which is called vertical scaling. Vertical scaling should be used as much as possible to accommodate user traffic increment immediately, and horizontal scaling is only performed when vertical scaling cannot allocate sufficient resource to meet the demand. We find that just to reserve redundant resource is far from sufficient to enable vertical scaling with good performance, and VNF placement plays a crucial role. In this paper, we model the scalable VNF placement problem formally, and prove that the problem is NP-hard. Then, we develop two online heuristic algorithms gradually to solve the problem. The sVNFP algorithm focuses on maximizing the vertical scalability, while the sVNFP-adv algorithm considers path length and server utilization concurrently. Simulation results show that our algorithms can reduce packet loss ratio by 15% to 40%, with a short path length and a high server utilization.