Modern clusters rely on centralized control planes, but decisions can be slow and fragile under control-plane degradations. We present Reflex, a bi-modal recovery design for clusters under control-plane degradation. Reflex adds a reflex-arc-like path that makes rapid takeover decisions from preprocessed local priorities while suppressing contention via lightweight coordination. After services are runnable, it performs steady-state reconstruction. Our evaluation shows bounded latency and robust conflict suppression under control-plane degradation and bursty failures.

Multi-agent workflows often generate repeated action requests when specialized agents interact with shared runtime objects. When requests target the same object under explicit compatibility conditions, a single execution result can serve multiple agents. This paper presents ActShare, a pre-execution coordination layer for reusable action requests. Before invoking a tool or a model, each agent emits a schema-constrained structured action request. ActShare compares the request against an active action table and a recent result cache. A request is attached only to a compatible owner action; otherwise, it is executed independently. When the owner action completes, ActShare writes the result once and dispatches it to all attached requests. We prototype ActShare in a stateful multi-agent debugging workflow and evaluate effect on repeated action reduction, token and tool-call savings.

Modern networks increasingly suffer from gray failures that are too fine-grained and short-lived for global re-optimization, while existing local mechanisms lack the context and control boundaries needed for effective bounded response. We propose CLEX, a contract-bounded local execution framework in which the controller defines per-device action boundaries and the device-side execution agent combines local semantics, runtime signals, and short-lived memory to select bounded local policy states that are realized through the local execution substrate. Implementation shows that CLEX enables agile local responses while reducing unnecessary action oscillation.

This paper proposes an optimization model to derive a resilient virtual network function (VNF) allocation aiming to maximize the number of accepted requests with considering VNF diversity and ensuring the requirements of node fault tolerances in a dynamic scenario, where the requests have random requirements, arriving and releasing time. The model considers fault tolerance assurance and satisfies the service requirements under different error patterns. The allocation provided by the proposed model ensures the required amount of processing ability in the situation that there are several failed nodes. The node fault tolerance can be set variously for different requirements of services. The proposed model selects and instantiates suitable replicas from the pools of replicas, and then determines the locations of these replicas instances. We develop a reinforcement learning approach for solving the proposed model including the design of the learning environment and the reward shaping. The numerical results show that the proposed model increases the number of accepted requests with ensuring the resiliency of the functions compared with baseline models in the examined cases, where the allocation of a request can be determined in tens of milliseconds.

Prompt function deployment and management is a key role in network function virtualization to improve the continuity and reliability of network services. Kubernetes is a system to deploy and manage functions automatically. Existing tools in Kubernetes do not provide automatic function deployment and management in a real-time and optimal manner. It does not provide a resource type to manage the migratable resource, either. This paper designs and implements a two-layer controller structure in Kubernetes to achieve the function deployment in a limited computation time with considering resource migration for allocation optimality. A controller in the lower layer manages the Pods for an intermediate allocation with a model or a heuristic algorithm to respond to requests promptly. A controller in the upper layer manages instances by optimizing resource allocations with considering resource migration; it maintains the Pods by keeping the current state (intermediate allocation) consistent with the desired state (optimal allocation). Our demonstration validates that the controller automatically manages the resources promptly and correctly.

This paper proposes an optimization model to derive the resilient virtual network function allocation in service function chains aiming to reduce the recovery time during the migrations from the primary functions to backup functions. We consider k-fault-tolerance assurance and satisfy the service requirements under different error patterns in this model. The allocation provided by the proposed model ensures that the processing ability satisfies the requirements even though there are k failed nodes in the network. Diversity splits a single VNF into a pool of replicas with different specifications. The diversity of both primary and backup functions are considered. Redundancy is used for recovering the failed functions. We formulate the proposed model as a mixed integer linear programming problem to select suitable replicas from the pools of replicas and decide the locations of these replicas for both primary and backup functions. The objective of the proposed model is to minimize the sum of the maximum recovery time among functions under all possible failure patterns which have k node failures. The numerical results show that the proposed model reduces the recovery times of VNFs with ensuring the resiliency of the functions compared with baseline models in the examined cases. We give some methods to improve the maximum resiliency at last.

Resource allocation and management is a key role in network function virtualization to improve the reliability of network services. Kubernetes is a system to deploy and manage the virtual network functions automatically. Existing tools in Kubernetes does not provide a resource type to define the backup Pods. It does not provide automatic resource management based on the user requests for the backup Pods, either. This paper designs and implements a custom resource and the corresponding controller in Kubernetes to manage the primary and backup resources of network functions. The custom resource is a set of Pods with different types, which includes primary, hot backup, and cold backup Pods. The controller manages the set of Pods and maintains the current state of the different types of Pods to keep the current state consistent with the desired state of each type of Pod. Demonstration validates that the controller automatically manage the primary and backups resources correctly.

Diversity in network function virtualization is to use a group of thin replicas to provide the network services under the required processing ability. Redundancy is to provide a certain number of replicas against function failures and improve network reliability. Kubernetes is a system to deploy and manage virtual network functions automatically. Existing tools in Kubernetes do not provide a resource type to provide required functions jointly considering VNF diversity and redundancy. This paper designs and implements a custom resource and the corresponding controller in Kubernetes to manage the VNF diversity and redundancy jointly. The controller selects suitable replicas from a pool of replica templates to satisfy the required processing ability with the minimum required number of replicas and converts the backup functions to the primary functions when the primary functions cannot provide the required ability. Demonstration validates that the controller automatically manages the resources correctly, improves the resource utilization, and increases the number of acceptable requests.

The availability schedule provides information on whether each network node is available at each time slot. The service interruptions caused by node unavailability marked in availability schedule can be suppressed if the functions are allocated according to the availability schedule. However, the given availability schedule may have gaps with the actual one and influence the VNF allocation. This paper proposes a robust optimization model to allocate virtual network functions (VNFs) in service function chains (SFCs) for time slots in sequence aiming to maximize the continuous available time of SFCs in a network with uncertain availability schedules by suppressing the interruptions caused by node unavailability marked in availability schedule and function reallocation. We formulate the problem as a mixed integer linear programming (MILP) problem over the given uncertainty set of the start time slot and period of unavailability on each node in the availability schedule. For solving the model in a practical time in a relative large size of network, we develop a heuristic algorithm. The numerical results show that the proposed model outperforms the baseline models under different levels of robustness in terms of the worst-case minimum number of the longest continuous available time slot in each SFC. The heuristic algorithm reduces the computation time with limited performance loss compared with the MILP approach. In the discussion, we introduce a constraint condition for the maintenance ability, which reduces the size of uncertainty set, and an extension for supporting more than one unavailability periods in the availability schedule on each node.

A suitable virtual network function (VNF) placement considering a node availability schedule extends service continuous serviceable time by suppressing service interruptions caused by function reallocation and node unavailabilities. However, function placement cannot avoid service interruptions caused by node unavailabilities. This paper proposes a primary and backup VNF placement model to avoid service interruptions caused by node unavailabilities by using backup functions. The considered backup functions have a period of startup time for preparation before they can be used and the number of them is limited. The proposed model is formulated as an integer linear programming problem to place the primary and backup VNFs based on the availability schedule at continuous time slots. We aim to maximize the minimum number of continuously available time slots in all service function chains (SFCs) over the deterministic availability schedule. We obverse that the proposed model considering the limited number of backup functions outperforms baseline models in terms of the minimum number of longest continuous available time slots in all SFCs. We introduce an algorithm to estimate the number of key unavailabilities at each time slot, which can find the unavailable nodes which are the bottlenecks to increase the service continuous available time at each time slot.

This paper proposes an optimization model to derive the virtual network function (VNF) allocation of time slots in sequence aiming to maximize the continuous available time of service function chains (SFCs) in a network. The proposed model suppresses service interruptions otherwise created by the unavailability of virtual machines (VMs) and the reallocation of VNFs. The proposed model computes VNF allocation in a series of time slots based on a VM availability schedule, which provides information on the availability of each VM in each time slot. We formulate the proposed model as an integer linear programming (ILP) problem with the goal of maximizing the minimum number of longest continuous available time slots in each SFC. We prove that the decision version of the VNF allocation problem (VNFA) is NP-complete. As the size of ILP problem increases, the problem is difficult to solve in a practical time. We develop a heuristic algorithm to solve the VNFA problem. Numerical results show that the proposed model improves the continuous available time of SFCs compared with existing models, which partially consider VM unavailability or VNF reallocation. We observe that the proposed model together with a consideration of routing reduces the path length of requests. The developed heuristic algorithm is faster than the ILP approach with a limited performance penalty.

A virtual network function allocation model to maximize continuous available time of service function chains was introduced in our previous work. The performance of this model needs to be evaluated on network devices. It is time-consuming and costly to deploy functions with real network devices. Existing simulation tools require powerful computation capability, which limits the usable cases. We implement a network service header based service function chain application which can be cooperated with the model. Demonstration validates that the application allocates functions by using the allocation from the model automatically and runs service function chains correctly.