Real-time applications require low latency and strict event ordering to ensure seamless operation. Distributed server processing is effective for this purpose, and there are two synchronization algorithms: a conservative synchronization algorithm (CSA) and an optimistic synchronization algorithm (OSA). OSA improves delay performance compared to CSA. While prior studies have considered OSA, they have not incorporated the impact of server failures. This paper proposes an OSA-based server allocation model for delay-sensitive applications with preventive start-time optimization (PreSO) under single-server failures (OSA-PreSO). The proposed OSA-PreSO model minimizes the largest total delay across all failure scenarios while satisfying constraints in OSA with PreSO under single-server failures. We formulate the proposed model as an integer linear programming (ILP) problem. In OSA-PreSO, the objective is to minimize the largest total delay across all failure scenarios, without giving special consideration to the total delay in the no-failure scenario. As a result, a penalty arises in the form of an increased total delay in the no-failure scenario. To reduce the penalty, we develop an improved OSA-PreSO model, OSA-PreSO-LP (low-penalty), which reduces the total delay in the nofailure scenario while maintaining the same delay characteristics in failure scenarios. We prove that the decision version of OSA-PreSO is NP-complete. We introduce heuristic algorithms to handle large-scale problems. Numerical results show that the proposed OSA-PreSO model reduces the delay compared to the conventional CSA-based model by effectively utilizing server memory resources. We observe that the proposed model achieves a lower largest total delay than start-time optimization and provides greater stability by preventing unnecessary user reassignments compared to run-time optimization. Numerical results also show that OSA-PreSO-LP reduces the penalty at most by 83%, while maintaining the same delay characteristics in failure scenarios.

Real-time applications require low latency and event order guarantees. Distributed server processing is effective for this purpose, and data consistency between servers is crucial. Although existing models in previous work handle data consistency, they do not address server failures. This paper proposes a server allocation model for a consistency-aware multi-server network for delay-sensitive applications with preventive start-time optimization (PSO) under single-server failures. The proposed model considers data consistency between servers and handles single-server failures with PSO. PSO determines the assignment to minimize the worst-case delay over all possible failure scenarios while avoiding service disruption for users connected to non-failed servers. We formulate the proposed model as an integer linear programming (ILP) problem. The decision version of the server allocation problem is proven to be NP-complete, and it becomes difficult to solve in a practical time when the problem size is large. We develop two polynomial-time approximation algorithms with theoretical performance analysis. Numerical results show that the proposed model outperforms start-time optimization in terms of the largest total delay and run-time optimization in terms of avoiding instability. The results also show that the faster of our two developed algorithms achieves a speedup ranging from 2.26×103to 4.37×106times compared to the ILP approach, while the maximum delay is, on average, only 1.029 times the optimal value. The results indicate that the speedup effect becomes more significant as the number of users and servers increases.

Real-time applications require low latency and event order guarantees. While distributed server processing is effective, data consistency across servers is crucial. Existing models address consistency but overlook server failures. This paper proposes a server allocation model for a consistency-aware multi-server network for delay-sensitive applications with preventive start-time optimization (PSO) under single-server failures. PSO determines the assignment to minimize the worst-case delay over all possible failure scenarios while avoiding service disruption for users connected to non-failed servers. We formulate the proposed model as an integer linear programming (ILP) problem. The decision version of the server allocation problem is proven to be NP-complete, and it becomes difficult to solve in a practical time when the problem size is large. We develop a polynomial-time approximation algorithm with theoretical performance analysis. Numerical results show that the proposed model outperforms start-time optimization in terms of the largest total delay and run-time optimization in terms of avoiding instability. Numerical results also show that our developed algorithm achieves a maximum speedup of 33.8 times compared to the ILP approach, while the maximum delay is, on average, only 1.016 times the optimal value.

Internet-of-Things (IoT) services require high performance regarding low delay and fault tolerance. Distributed server allocation is well-suited for meeting these requirements in IoT monitoring services. Previous work focused on reducing delay but overlooked the need for fault tolerance in distributed server allocation. This paper proposes a distributed server allocation model based on preventive start-time optimization (PSO) for IoT monitoring services against server failure. The proposed model preventively determines the server allocation to minimize the largest maximum delay between IoT devices and application servers and between database and application servers among all failure patterns. We formulate the proposed model as an integer linear programming (ILP) problem. We introduce a server allocation algorithm based on PSO to accelerate the computation to obtain an optimal server allocation, compared to the ILP approach. We prove that the introduced algorithm obtains a PSO-based optimal allocation in polynomial time. Numerical results show that the introduced algorithm outputs an optimal server allocation faster than the ILP approach. We compare the PSO-based server allocation with allocations based on the start-time and run-time optimization. We observe that the PSO-based allocation reduces the largest maximum delay by 5.5% for a network model with eleven servers compared to the start-time optimization and avoids unnecessary network disconnections while increasing the maximum delay by 5.1% compared to the run-time optimization.

Delay-sensitive applications have been provided through a low-delay network utilizing multiple edge clouds. For applications that involve sharing status among multiple users, it is crucial to prevent longer communication delays for users who are farther from the application server compared to those who are closer. To address this issue, this paper proposes a distributed processing network design scheme for virtual processing platforms using low-delay networks and widely distributed servers. The proposed scheme introduces Tapl as a given parameter for correcting events in occurrence order. Events within Tapl delay are sorted in occurrence order. The proposed scheme can change its operation mode from a conservative synchronization to an optimistic synchronization depending on the setting of Tapl. The proposed scheme is formulated as a mixed-integer linear programming problem to determine users’ and servers’ distributed processing network configuration. We evaluate the proposed scheme on two different network topologies. Numerical results indicate that, depending on the setting of Tapl, the proposed scheme can reduce the maximum amount of memory used for rollback processes in optimistic synchronization-based applications or realize a conservative synchronization algorithm. The computation time under the condition of 1000 users is within a maximum of nine sec, an acceptable amount of time for preparation before starting a planned service. These results indicate that the proposed scheme realizes event order correction with excellent delay characteristics and applies to virtual processing platforms.

In real-time applications under wide-area networks, providing a demanded quality of service for end users is an issue. Recent studies adopt distributed processing for server selection problems to reduce data synchronization delay and total interaction delay, assuming that link delays over the distributed system are exactly known. No study has addressed the problem of such a distributed server selection in properly handling the delay uncertainty. This paper proposes a robust optimization model for the distributed server selection problem against the delay uncertainty. We handle the delay uncertainty of user-server and server-server links by defining two -ellipsoidal uncertainty sets. The proposed model determines allocated servers for multiple users to minimize the weighted sum of data synchronization delay and total interaction delay over the distributed system. We formulate the proposed model as a mixed integer second-order cone programming problem. We prove that the distributed server selection problem with uncertain delays is NP-complete. We compare the proposed model with baseline models, focusing on delay uncertainty and distributed processing. The numerical results show that the proposed model can achieve a lower objective value than the baseline models, indicating the benefit of utilizing -ellipsoidal uncertainty sets to handle delay uncertainty.

Low-delay networking and edge computing will enable mission-critical applications to be delivered over wide-area networks. We consider this trend to be the realization that all users can share an application space without feeling any distance difference. We propose a distributed processing scheme that keeps the order of event occurrence regardless of the distance between users and an application server. The proposed scheme can be applied to both optimistic synchronization algorithms (OSA) and conservative synchronization algorithms (CSA). In the proposed scheme, arrival events with a delay within a predefined set time (correction time) are sorted in order of occurrence before application processing. We formulate the proposed scheme as an integer linear programming (ILP) problem. The objective function of ILP consists of the number of users excluded from the delay quality, the amount of memory consumed for a rollback in OSA, and the maximum end-to-end delay. The three parts of the objective function are set weight and the sum of parts with weight is minimized. We evaluate the proposed scheme for 1000 users distributed in two types of network models. Numerical results indicate that the proposed scheme reduces memory consumption compared to that of the conventional OSA scheme. The proposed scheme works as CSA in which all events are sorted in the occurrence order if the correction time is set above the delay for the slowest event to arrive at the server.

This paper proposes a polynomial-time algorithm for a server allocation problem in delay-sensitive Internet-of-Things (IoT) monitoring services. The server allocation problem determines the appropriate servers to which the database and application are allocated to minimize the maximum delay between the latest update of reference data and the start of application processing for monitoring data . The server allocation problem was previously handled by expressing it as an integer linear programming (ILP) problem. Nevertheless, it fails to meet the computational time complexity needed to solve the problem, and it does not offer a more efficient technique than the ILP approach. The proposed algorithm consists of two components. The first step entails choosing utilization servers for both the database and the application. Next, the second phase entails matching each usage server and its corresponding IoT device . We prove that the proposed algorithm obtains an optimal solution in polynomial time . We compare computation times between the ILP approach and the proposed algorithm. Numerical results show that the proposed algorithm obtains the optimal solution faster than the ILP approach.

When mission-critical applications are provided over a network, high availability is required in addition to a low delay network. This paper proposes a multi-homing network design model, named MHND, to balance a low delay and high availability when distributed processing applications use multiple processing servers. MHND maintains the event occurrence order with a multi-homing configuration using conservative synchronization. We formulate MHND as an integer linear programming problem to minimize the delay. We prove that the distributed server allocation problem with MHND is NP-complete. Numerical results indicate that, as a multi-homing number, which is the number of servers to which each user belongs, increases, the availability increases while increasing the delay. Two or more multi-homing can achieve approximately an order of magnitude higher availability compared to that of the conventional single-homing at the expense of a delay increase of 1.25 times. By using MHND, flexible network design is achieved based on the acceptable delay in service and the required availability.

This paper proposes a network design approach considering data consistency for a delay-sensitive distributed processing system. The data consistency is determined by collating the own state and the two states of slave servers. If the state is mismatched with other servers, the rollback process is initiated to modify the state to guarantee data consistency. In the proposed approach, the select servers and the master-slave server pairs are determined to minimize the end-to-end delay and the delay for data consistency. We formulate the proposed approach as an integer linear programming problem. We evaluate the delay performance and computation time. The proposed approach reduces the delay for data consistency by 6.8-31.2% compared to that of a typical approach that collates the status of all servers at the master server. The computation time is a few seconds, which is an acceptable time for network design before service launch. These results indicate that the proposed approach is effective for delay-sensitive applications.

This paper proposes an optimistic synchronization algorithm (OSA) based server selection scheme with successive participation scenario. In the scenario, we introduce a participating-domain segmentation and determe recommended servers before user participation. Numerical results indicate that the proposed scheme reduces the latency compared to the non-domain segmentation approach (conventional scheme) and overcomes latency fluctuation.

In distributed processing for communication services, a proper server selection scheme is required to suppress delay by ensuring the event occurrence order. Although a conservative synchronization algorithm (CSA) has been used in this issue, an optimistic synchronization algorithm (OSA) can be a potential candidate for synchronizing distributed systems. In comparison with CSA, which reproduces events in occurrence order before processing application, OSA can be feasible to realize low delay communication as the processing events arrive sequentially. This paper proposes an optimal server selection scheme considering OSA for distributed processing systems to minimize end-to-end delay under the condition that the holding time for application status is limited. In other words, the end-to-end delay is minimized based on the allowed rollback time for application design or quality-of-service. Numerical results indicate that the delay of the proposed scheme can be reduced by up to a quarter compared to that of the conventional scheme that is based on CSA.

In the modern era, we need to deploy several functionalities either on central cloud servers or edge cloud servers to provide Internet of Things (IoT)-based services via a wide-area network. Typically a huge database and several non-real-time functions are deployed on the central cloud servers. On the other hand, real-time functions are deployed on edge cloud servers. For delay-sensitive services, this approach has an issue in completing the analysis with the latest information when the delay between the central and edge cloud servers is large. In this paper, we propose an allocation scheme of database and applications, which minimizes the delay from the latest update of the database to analyze the real-time data from IoT devices. In the proposed scheme, the delay is minimized considering two constraints of each server, which are maximum accommodating capacity of the IoT devices and whether the database function can be deployed. Numerical results observe that the proposed scheme reduces the delay of analysis compared to the conventional scheme. These results indicate that the proposed scheme can configure a low-delay network for delay-sensitive IoT services with data analysis.

This paper proposes a distributed server allocation model with the preventive start-time optimization against a single server failure. The proposed model preventively determines the assignment of servers to users under each failure pattern to minimize the largest maximum delay among all failure patterns. We formulate the proposed model as an integer linear programming problem. We prove the NP-completeness for the considered problem. The numerical results reveal that the proposed model reduces the largest maximum delay compared to one baseline; it avoids instability caused by the unnecessary disconnection, which frequently occurs in the other baseline.

This paper proposes a server selection scheme in successive participation scenario based on the segmentation of participating-domain to suppress the latency. In the proposed scheme, the users participate for server selection one after another. The proposed scheme determines a recommended server, and a new participating user selects the recommended server first. Recommended servers are determined in advance at the condition that participating users exist in all regions in users' participation domain. A recommended server is determined for each divided region to minimize the latency when there exists one user at each region. The new participating user selects the recommended available server, where the user is located. We formulate an integer linear programming problem to determine the recommended servers for the proposed scheme. We use the outcome of the recommended server finding process as the input for server selection. Numerical results express that smaller latency is obtained using the proposed scheme compared to the conventional greedy based server selection scheme, by employing some additional computations for finding the recommended servers before participating of users.

The performance of packet processing applications is dependent on memory accesses speed of network systems. Table lookup requires fast memory accesses and is one of the most common processes in various packet processing applications, which can be a dominant performance bottleneck. Therefore, in Network Function Virtualization (NFV)-aware environment, on-chip fast cache memories of a CPU of general-purpose hardware become critical to achieve high performance packet processing over tens of Gbps. In addition, multiple types of applications and complex applications are executed in the same system simultaneously in carrier network systems, which require the capacity of cache memories as well. In this paper, we propose a packet processing architecture that utilizes interleaved 3 Dimensional (3D)-stacked Dynamic Random Access Memory (DRAM) devices as off-chip Last Level Cache (LLC) in addition to several levels of dedicated cache memories of each CPU core. Entries of a lookup table are distributed in every bank and vaults to utilize both bank interleaving and vault-level memory access parallelism. Frequently accessed entries in 3D-stacked DRAM are also cached in dedicated on-chip cache memories of each CPU core. The evaluation results show that the proposed architecture reduces the memory access latency by 57 % and increases the throughput by 100 % with reducing blocking probability about 10 % compared to the conventional architecture with common on-chip LLC. These results indicate that 3D-stacked DRAM can be practical as off-chip LLC in parallel packet processing running on multiple CPU cores simultaneously.

Emergence of new network services such as Internet of Things (IoT) and edge computing accelerates the increase of traffic volume, the number of connected devices and the diversity of communication. Next generation carrier network infrastructure should be much more scalable and adaptive to rapid increase and divergence of network demand with much lower cost. More virtualization-aware, flexible and inexpensive system based on general-purpose hardware is necessary to transform traditional carrier network into more adaptive, next generation network. In this paper, we propose a carrier-scale packet processing system which utilizes 3 Dimensional (3D)-stacked Dynamic Random Access Memory (DRAM) device. The proposed system augments memory access concurrency by leveraging vault-level parallelism and bank interleaving of 3D-stacked DRAM. The system uses hash-function-based distributor of memory requests to each set of vault and bank which accommodates a portion of original carrier-scale huge tables. We introduce an analytical model for the system. The evaluation result shows that our proposed system can achieve more than 100 Gbps in carrier-scale packet processing where main memory accesses are inevitable since tiny CPU cache memory is insufficient to accommodate huge tables. Our analytical model is independent of specification of a particular device, which can be applied to any DRAM systems.

Due to high-density implementation, large-capacity routers are very heavy and consume a huge amount of power. Such routers require the capability for power-supply and air-cooling equipment to be increased. Upon introducing such routers, the physical limitations of the infrastructure should be considered. In addition, wider space in machine rooms to maintain reliability criteria is required, even though space is a precious resource and the purpose of high-density routers is to save floor space. We describe earthquake resistance of high weight, power-supply stability of high power-supply current, and air-cooling reliability of high power consumption. We propose solutions for improved cabinets and power-supply devices and the allocation of this equipment to satisfy network reliability, such as earthquake resistance, power-supply stability, and continuous duration after air-cooling stops. This leads to saving floor area space.

This paper analyzes the computational time complexity of the allocation problem for data processing functions among multiple users and distributed servers in the distributed processing communication scheme for a real-time network application. In the distributed processing communication scheme, the application is processed on a data processing function in the distributed servers in order to minimize the delay time. We prove that the allocation problem for data processing functions among multiple users and distributed servers is an NP-complete problem.