Geo-distributed Data Centers Research Articles

Service Function Chain Deployment and Network Flow Scheduling in Geo-Distributed Data Centers

Network Function Virtualization (NFV), as an emerging solution to virtualize network services traditionally running on proprietary, dedicated devices, can effectively reduce the cost of big data processing service providers and improve Quality of Service (QoS) by running a chain of ordered Virtual Network Functions (VNFs) on commodity hardware. One fundamental and critical problem of big data processing with NFV is how to deploy the chained VNFs and dispatch corresponding network flows to process the big data traffics so that the service cost can be minimized with guaranteed QoS. In this paper, we study the problem of VNF deployment and flow scheduling in distributed data centers with joint consideration of the service requirements and the resource capacity, and prove its NP-hardness through reduction from the k-level uncapacitated facility location problem. A two-phased algorithm is also devised by first balancing VNF resource requirements and then selecting VNF locations with a low complexity of $O(M^2\log _2 M)$ and an approximation ratio of $K + \rho$ . Finally, with extensive simulation experiments, the result shows that our algorithm can efficiently reduce the total cost of VNF deployment and flow communication in various scenarios.

A Framework of Hypergraph-Based Data Placement Among Geo-Distributed Datacenters

Data-intensive applications need to address the problem of properly placing the set of data items in geo-distributed storage nodes. Traditional techniques use the hashing method to achieve the load balance among nodes such as those used in Hadoop and Cassandra, but are not efficient for the requests reading multiple data items in one transaction, especially when the source locations of requests are also distributed. Some recent papers proposed the managed data placement schemes for online social networks, but have a limited scope of applications due to their focuses. We propose a general hypergraph-based data placement framework, which considers both the performance metrics related to the co-location of associated data and those related to the exact location of fulfilling each requested data item. In the framework, we present the methods to convert the optimization objectives into hypergraph models and employ a hypergraph partitioning to efficiently partition the set of data items and place them in distributed nodes. Further, we extend the scheme into replica placement where we need to find multiple locations to place the replicas of the same data item. Through extensive experiments based on trace-based datasets, we evaluate the performance of the proposed framework and demonstrate its effectiveness.

Optimizing Geo-Distributed Data Analytics with Coordinated Task Scheduling and Routing

Recent trends show that cloud computing is growing to span more and more globally distributed datacenters. For geo-distributed datacenters, there is an increasingly need for scheduling algorithms to place tasks across datacenters, by jointly considering WAN traffic and computation. This scheduling must deal with situations such as wide-area distributed data, data sharing, WAN bandwidth costs and datacenter capacity limits, while also minimizing makespan. However, this scheduling problem is NP-hard. We propose a new resource allocation algorithm called HPS + , an extension to Hypergraph Partition-based Scheduling. HPS + models the combined task-data dependencies and data-datacenter dependencies as an augmented hypergraph, and adopts an improved hypergraph partition technique to minimize WAN traffic. It further uses a coordination mechanism to allocate network resources closely following the guidelines of task requirements, for minimizing the makespan. Evaluation across the real China-Astronomy-Cloud model and Google datacenter model show that HPS + saves the amount of data transfers by upto 53 percent and reduces the makespan by 39 percent compared to existing algorithms.

Resource and Cost Aware Glowworm Mapreduce Optimization Based Big Data Processing in Geo Distributed Data Center

Handling large data in geographically distributed information centers with resource and cost optimization is a key challenge. With several approaches being designed, handling a large volume of data in multiple datacenters in an inappropriate manner yet is considered to be a time-consuming process. To address these issues, a Multivariate Metaphor based Metaheuristic Glowworm Swarm Map-Reduce Optimization (MM-MGSMO) technique is presented. Here, with search space and large data volume as input for geo-distributed datacenters, glowworm (i.e. virtual machine) population is initialized. With each glowworm possessing a certain amount of luciferin (i.e. objective function), multiple objective functions (i.e. bandwidth, storage capacity, energy and computation cost) are defined for each virtual machine. Next, the glowworm position is updated according to the neighboring factor by means of probability. Followed by this, MapReduce function identifies the optimal virtual machine and accordingly allocation is performed, therefore improving data allocation efficiency. Besides, the workload is assigned across datacenters, reduction in computation cost and storage capacity is guaranteed. Experimental evaluation of MM-MGSMO approach with existing methods attained improved performances with factors such as data allocation efficiency, false-positive rate, storage capacity compared with other cutting edge technologies such as Joint optimization algorithm and Game theory-based dynamic resource allocation strategy.

Distributed Data Strategies to Support Large-Scale Data Analysis Across Geo-Distributed Data Centers

As the volume of data grows rapidly, storing big data in a single data center is no longer feasible. Hence, companies have developed two scenarios to store their big data in multiple data centers. In the first scenario, the company's big data are distributed in multiple data centers without data replication. In the second scenario, data are also stored in multiple data centers but important data are replicated in these data centers to increase data safety and availability. However, in these scenarios, analyzing big data distributed in multiple data centers becomes a challenging task. In this paper, we propose two data distribution strategies to support big data analysis across geo-distributed data centers. In these strategies, we use the recent Random Sample Partition data model to convert big data into sets of random sample data blocks and distribute these data blocks into multiple data centers either without replication or with replication. In analyzing big data in multiple data centers without replication, we randomly select samples of data blocks from multiple data centers and download the sample data blocks to one data center for analysis. In the second strategy with replication of data blocks, we can analyze big data on any data center by randomly selecting a sample of data blocks replicated from other data centers. This strategy avoids data transformation between data centers. We demonstrate the performance of the two strategies in big data analysis by using simulation results produced on one local data center and four AWS data centers in North America, Asia, and Australia.

Renewable Energy-Aware Big Data Analytics in Geo-Distributed Data Centers with Reinforcement Learning

In the age of big data, companies tend to deploy their services in data centers rather than their own servers. The demands of big data analytics grow significantly, which leads to an extremely high electricity consumption at data centers. In this paper, we investigate the cost minimization problem of big data analytics on geo-distributed data centers connected to renewable energy sources with unpredictable capacity. To solve this problem, we propose a Reinforcement Learning (RL) based job scheduling algorithm by combining RL with neural network (NN). Moreover, two techniques are developed to enhance the performance of our proposal. Specifically, Random Pool Sampling (RPS) is proposed to retrain the NN via accumulated training data, and a novel Unidirectional Bridge Network (UBN) structure is designed for further enhancing the training speed by using the historical knowledge stored in the trained NN. Experiment results on real Google cluster traces and electricity price from Energy Information Administration show that our approach is able to reduce the data centers' cost significantly compared with other benchmark algorithms.

Eco-Friendly Powering and Delay-Aware Task Scheduling in Geo-Distributed Edge-Cloud System: A Two-Timescale Framework

Power consumption and task latency are two crucial issues in edge-cloud computing. This paper mainly aims to promote the use of clean power in geo-distributed data centers (DCs) in a deregulated electricity market where customers are allowed to buy power from multiple suppliers, combined with the guarantee of task latency. To alleviate the conflict between frequent switches of servers and the uncertainty of task arrivals in DCs, this paper proposes a two-timescale framework consisting of the long-term capacity planning of geo-distributed DCs and the real-time task dispatching from edge gateways (EGs) to DCs. First, DCs make long-term plans on the number of active servers aiming at the eco-friendly and delay-aware power cost minimization, which is formulated as problem P. Specifically, we introduce a convex pollution indicator function (PIF) to measure the pollution cost of the various types of powers sold by different suppliers, which can encourage the use of cleaner power and improve power savings. Second, in each sub-slot, each EG separately optimizes its individual mixed strategies of task dispatching to DCs with the knowledge of the planned capacities and the real-time queue backlogs of DCs, where a Lyapunov optimization framework is applied. Finally, we give the corresponding distributed algorithm design. Simulation results reveal that our method can realize the trade-off between the power cost and the delay cost of requests, and improve the clean power usage by up to 50%-60% of the total power usage in DCs. Additionally, comparisons with other schemes show that our approach can provide more stable guarantees of task latency in different situations of workload density, which benefits from the diverse-timescale optimizations of capacities of DCs and task routing from EGs to DCs.

Bulk Savings for Bulk Transfers: Minimizing the Energy-Cost for Geo-Distributed Data Centers

With the fast proliferation of cloud computing, major cloud service providers, e.g., Amazon, Google, Facebook, etc., have been deploying more and more geographically distributed data centers to provide customers with better reliability and quality of services. A basic demand in such a geo-distributed data center system is to transfer bulk volumes of data from one data center to another. Geographic distribution and large delay-tolerance of such inter-data-center bulk data transfers provide cloud service providers opportunities to optimize the operating cost. Most existing studies on inter-data-center bulk data transfers focus on minimizing the network bandwidth cost. However, the energy-cost of the bulk data transfers, which also accounts for a large proportion of operating cost in the data centers, still remains unexplored. This is an important problem, especially in the multi-electricity-market environment, where the electricity price exhibits both spatial and temporal diversities. In this paper, we systematically study the problem of how to route and schedule inter-data-center bulk data transfers to minimize the energy-cost for geo-distributed data centers. We model this problem as a min-cost multi-commodity flow problem and develop an efficient two-stage optimization method to solve it. Extensive evaluations with real-life inter-data-center network and electricity prices show that our method brings significant energy-cost savings over existing bulk data transfer methods.

A Hybrid Learning Framework for Service Function Chaining Across Geo-Distributed Data Centers

Service function chaining (SFC) focuses mainly on deploying various network functions in geographically distributed data centers and providing interconnect routing among them. Traditional (convex optimization-based) SFC algorithms exhibit some disadvantages on the scalability and accuracy. Recently, researches have shown the effectiveness of deep reinforcement learning (DRL) in the field of SFC. However, current DRL-based algorithms possess an extremely large action space, which leads to slow convergence and poor scalability. Some researchers relieve this issue by reformulating the SFC problem, which usually results in low utilization and high cost. To address this issue, we develop a hybrid DRL-based framework which decouples the VNF deployment and flow routing into different modules. In the proposed framework, a DRL agent is only responsible for learning the policy of VNF deployment. We customize the structure of the agent base on deep deterministic policy gradient (DDPG) and adopt several techniques to improve the learning efficiency, such as adaptive parameter noise, wolpertinger policy, and prioritized experience replay. The flow routing is conducted in a game-based module (GBM). We design a decentralized routing algorithm for the GBM to address the scalability. The end-to-end latency of flows is minimized while the resource capacity and location constraints are satisfied. During the learning process of the proposed framework, the DRL agent improves its deployment policy with the reward from the GBM (the value of reward depends on flow routing). Thus, the VNF deployment and flow routing are still jointly optimized. Compared to existing DRL-learning algorithms, the proposed hybrid DRL framework can achieve a lower cost since 1) the action space is significantly reduced due to flow routing decoupling; 2) the flow routing procedure is more efficient (the GBM adopts model-based information, e.g., the gradient). Through trace-driven simulations, we show the efficiency of our algorithm compared to existing DRL-based algorithms.

Joint Online Coflow Optimization Across Geo-Distributed Datacenters

Growing data volumes have been generated, stored and processed across geographically distributed datacenters. Processing such geo-distributed datasets should consider task or job placement, heterogeneous available link bandwidth, and routing decisions. Thus optimizing the transmissions of inter-datacenter flows, especially coflows that capture application-level semantics, is important. However, prior solutions on coflow scheduling either merely ignore routing or fix the endpoint placement. This article focuses on the problem of jointly considering endpoint placement, coflow scheduling, and coflow routing to minimize the average coflow completion time (CCT) across geo-distributed datacenters. This paper first presents the system model and problem formulation, then derives a randomized approximation algorithm for a single coflow. After that it also proposes an online algorithm to handle the complex cases of multiple coflows. Through theoretical analysis, a proof that the proposed algorithms have a non-trivial competitive ratio is given. Results from extensive simulations demonstrate that the proposed algorithms can significantly reduce the CCT of coflows and at the same time, have similar algorithm run time, compared with the state-of-the-art solutions.

Mitigating Bottlenecks in Wide Area Data Analytics via Machine Learning

Over the past decade, we have witnessed exponential growth in the density (petabyte-level) and breadth (across geo-distributed datacenters) of data distribution. It becomes increasingly challenging but imperative to minimize the response times of data analytic queries over multiple geo-distributed datacenters. However, existing scheduling-based solutions have largely been motivated by pre-established mantras (e.g., bandwidth scarcity). Without data-driven insights into performance bottlenecks <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">at runtime</i> , schedulers might blindly assign tasks to workers that are suffering from unidentified bottlenecks. In this paper, we present <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Lube</i> , a system framework that minimizes query response times by detecting and mitigating bottlenecks at runtime. Lube monitors geo-distributed data analytic queries in real-time, detects potential bottlenecks, and mitigates them with a bottleneck-aware scheduling policy. Our preliminary experiments on a real-world prototype across Amazon EC2 regions have shown that Lube can detect bottlenecks with over 90 percent accuracy, and reduce the median query response time by up to 33 percent compared to Spark's built-in locality-based scheduler.

Workload Based Geo-Distributed Data Center Planning in Fast Developing Economies

Data centers (DCs), as the core infrastructure of cloud computing, are experiencing explosive growth worldwide. The planning of geo-distributed DCs has to take into account geographically differentiated costs, so that the overall expenditure could be minimized. In fast developing economies, the planning of DCs is further complicated by the fact that demands are quickly evolving, both in the type of services, and in geographical distributions. This paper is dedicated to the planning of DCs in fast developing economies. Instead of modeling demands by aggregated number of servers, we introduce a workload-based model to better capture the quickly changing nature of demand composition. The variability and dynamics of computational demand are modeled by dynamic compositions of workflows of different types. We provide approaches to solve the planning problem. Finally, we apply the model to China by using real-life data, and show how factors like economic growth, population migration and latency constraint may affect DC planning. Our research suggests that factors endemic in fast developing economies can significantly influence the overall cost and performance. For instance, by considering these factors, one can save 5.8% in cost in a five-year planning period, and 5% population migration may lead to 30ms increase in latency.

A Truthful and Efficient Incentive Mechanism for Demand Response in Green Datacenters

Datacenter demand response is envisioned as a promising tool for mitigating operational stability issues faced by smart grids. It enables significant potentials in peak load reduction and facilitates the incorporation of distributed generation. Monetary refund from the smart grid can also alleviate the cloud's burden in escalating electricity cost. However, the current demand response paradigm is inefficient towards incentivizing a cloud service provider (CSP) that operates geo-distributed datacenters. To incentivize CSP participation, this work presents an auction mechanism that enables smart grids to voluntarily submit bids to the CSP to procure diverse amounts of demand response with different payments. To maximize the social welfare of the auction, the CSP that acts as the auctioneer needs to solve the winner determination problem at large-scale. By applying the proximal Jacobian alternating direction method of multipliers, we propose a distributed algorithm for each datacenter to solve a small-scale problem in a parallel fashion. Desirable properties of the proposed auction, such as social welfare maximization and truthfulness are achieved through Vickrey-Clarke-Groves (VCG) payment. Through extensive evaluations based on real datacenter workload traces and IEEE 14-bus test systems, we demonstrate that our incentive mechanism constitutes a win-win mechanism for both the geo-distributed cloud and the smart grid.

Cost-Aware Partitioning for Efficient Large Graph Processing in Geo-Distributed Datacenters

Graph processing is an emerging computation model for a wide range of applications and graph partitioning is important for optimizing the cost and performance of graph processing jobs. Recently, many graph applications store their data on geo-distributed datacenters (DCs) to provide services worldwide with low latency. This raises new challenges to existing graph partitioning methods, due to the multi-level heterogeneities in network bandwidth and communication prices in geo-distributed DCs. In this article, we propose an efficient graph partitioning method named Geo-Cut, which takes both the cost and performance objectives into consideration for large graph processing in geo-distributed DCs. Geo-Cut adopts two optimization stages. First, we propose a cost-aware streaming heuristic and utilize the one-pass streaming graph partitioning method to quickly assign edges to different DCs while minimizing inter-DC data communication cost. Second, we propose two partition refinement heuristics which identify the performance bottlenecks of geo-distributed graph processing and refine the partitioning result obtained in the first stage to reduce the inter-DC data transfer time while satisfying the budget constraint. Geo-Cut can be also applied to partition dynamic graphs thanks to its lightweight runtime overhead. We evaluate the effectiveness and efficiency of Geo-Cut using real-world graphs with both real geo-distributed DCs and simulations. Evaluation results show that Geo-Cut can reduce the inter-DC data transfer time by up to 79 percent (42 percent as the median) and reduce the monetary cost by up to 75 percent (26 percent as the median) compared to state-of-the-art graph partitioning methods with a low overhead.

Efficient Data Placement and Replication for QoS-Aware Approximate Query Evaluation of Big Data Analytics

Enterprise users at different geographic locations generate large-volume data that is stored at different geographic datacenters. These users may also perform big data analytics on the stored data to identify valuable information in order to make strategic decisions. However, it is well known that performing big data analytics on data in geographical-located datacenters usually is time-consuming and costly. In some delay-sensitive applications, the query result may become useless if answering a query takes too long time. Instead, sometimes users may only be interested in timely approximate rather than exact query results. When such approximate query evaluation is the case, applications must sacrifice timeliness to get more accurate evaluation results or tolerate evaluation result with a guaranteed error bound obtained from analyzing the samples of the data to meet their stringent timeline. In this paper, we study quality-of-service (QoS)-aware data replication and placement for approximate query evaluation of big data analytics in a distributed cloud, where the original (source) data of a query is distributed at different geo-distributed datacenters. We focus on the problems of placing data samples of the source data at some strategic datacenters to meet stringent query delay requirements of users, by exploring a non-trivial trade-off between the cost of query evaluation and the error bound of the evaluation result. We first propose an approximation algorithm with a provable approximation ratio for a single approximate query. We then develop an efficient heuristic algorithm for evaluating a set of approximate queries with the aim to minimize the evaluation cost while meeting the delay requirements of these queries. We finally demonstrate the effectiveness and efficiency of the proposed algorithms through both experimental simulations and implementations in a real test-bed, real datasets are employed. Experimental results show that the proposed algorithms are promising.

Stackelberg Game for Energy-Aware Resource Allocation to Sustain Data Centers Using RES

Smart Grid (SG) has emerged as one of the most powerful technologies of the modern era for an efficient energy management by integrating information and communication technologies (ICT) in the existing infrastructure. Among various ICT, cloud computing (CC) has emerged as one of the leading service providers which uses geo-distributed data centers (DCs) to serve the requests of users in SG. In recent times, with an increase in service requests by end users for various resources, there has been an exponential increase in the number of servers deployed at various DCs. With an increase in the size, the energy consumption of DCs has increased many folds which leads to an increase in overall operational cost of DCs. However, efficient resource allocation among these geo-distributed DCs may play a vital role in reducing the energy consumption of DCs. Moreover, with an increase in harmful emissions, the use of renewable energy sources (RES) can benefit DCs, SG, and society at large. Keeping focus on these points, in this paper, an energy-aware resource allocation scheme is proposed using a Stackelberg game for energy management in cloud-based DCs. For this purpose, a cloud controller is used to receive the requests of users which then distributes these requests among geo-distributed DCs in such a way that the energy consumption of DCs is sustained by RES. However, if energy consumption of DCs is not sustained by RES then the energy is drawn from the grid. The requests of users are routed to the DC which is offered lowest energy tariff from the grid. For this purpose, a Stackelberg game for energy trading is also proposed to select the grid offering lowest energy tariff to DCs. The proposed scheme is evaluated using various performance metrics using Google workload traces. The results obtained show the effectiveness of the proposed scheme.

Privacy Regulation Aware Process Mapping in Geo-Distributed Cloud Data Centers

Recently, various applications including data analytics and machine learning have been developed for geo-distributed cloud data centers. For those applications, the ways of mapping parallel processes to physical nodes (i.e., “process mapping”) could significantly impact the performance of the applications because of non-uniform communication cost in geo-distributed environments. What's more, the different data privacy requirements in geo-distributed data centers pose additional constraints on process mapping solutions. While process mapping has been widely studied in grid/cluster environments, few of the existing studies have considered the problem in geo-distributed cloud environment, which is a challenging task due to the multi-level data privacy constraints, heterogeneous network performance and process failures. In this paper, we introduce the special privacy requirements in geo-distributed data centers and formulate the geo-distributed process mapping problem as an optimization problem with multiple constraints. We develop a new method to efficiently find good process mapping solutions to the problem. Experimental results on real clouds (including Amazon EC2 and Windows Azure) and simulations demonstrate that our proposed approach can achieve significant performance improvement compared to the state-of-the-art algorithms.

TailCutter: Wisely Cutting Tail Latency in Cloud CDNs Under Cost Constraints

Cloud computing platforms enable applications to offer low-latency services to users by deploying data storage in multiple geo-distributed data centers. In this paper, through benchmark measurements on Amazon AWS and Microsoft Azure together with an analysis of a large-scale dataset collected from a major cloud CDN provider, we identify the high tail latency problem in cloud CDNs, which can substantially undermine the efficacy of cloud CDNs. One crucial idea to reduce the tail latency is to send requests in parallel to multiple clouds in cloud CDNs. However, since application providers often have a budget for using cloud services, deciding how many chunks to download from each cloud and when to download chunks in a cost-efficient manner still remain as open problems in our concerned scenario. To address the problem, we present TailCutter, a workload scheduling framework that aims at optimizing the tail latency while meeting cost constraints given by application providers. Specifically, we formulate the tail latency minimization (TLM) problem in cloud CDNs and design the receding horizon control based maximum tail minimization algorithm (RHC-based MTMA) to efficiently solve the TLM problem in practice. We implement TailCutter across multiple data centers of Amazon AWS and Microsoft Azure. Extensive evaluations using a large-scale real-world data trace (collected from a major ISP) illustrate that TailCutter can reduce up to 58.9% of the 100th-percentile user-perceived latency, as compared with alternative solutions under the cost constraint.

Scheduling Jobs across Geo-Distributed Datacenters with Max-Min Fairness

It has become routine for large volumes of data to be generated, stored, and processed across geographically distributed datacenters. To run a single data analytic job on such geo-distributed data, recent research proposed to distribute its tasks across datacenters, considering both data locality and network bandwidth across datacenters. Yet, it remains an open problem in the more general case, where multiple analytic jobs need to fairly share the resources at these geo-distributed datacenters. In this paper, we focus on the problem of assigning tasks belonging to multiple jobs across datacenters, with the specific objective of achieving max-min fairness across jobs sharing these datacenters, in terms of their job completion times. We formulate this problem as a lexicographical minimization problem, which is challenging to solve in practice due to its inherent multi-objective and discrete nature. To address these challenges, we iteratively solve its single-objective subproblems, which can be transformed to equivalent linear programming (LP) problems to be efficiently solved, thanks to their favorable properties. As a highlight of this paper, we have designed and implemented our proposed solution as a fair job scheduler based on Apache Spark, a modern data processing framework. With extensive evaluations of our real-world implementation on Amazon EC2 and large-scale simulations, we have shown convincing evidence that max-min fairness has been achieved and the worst job completion time has been significantly improved using our new job scheduler.

Wide-Area Spark Streaming: Automated Routing and Batch Sizing

Modern stream processing frameworks, such as Spark Streaming, are designed to support a wide variety of stream processing applications, such as real-time data analytics in social networks. As the volume of data to be processed increases rapidly, there is a pressing need for processing them across multiple geo-distributed datacenters. However, these frameworks are not designed to take limited and varying inter-datacenter bandwidth into account, leading to longer query latencies. In this paper, we present the design and implementation of an extended Spark Streaming framework to automatically and optimally schedule tasks, select data flow routes and determine micro-batch sizes across geo-distributed datacenters in wide-area networks. To make these decisions, we propose a sparsity-regularized ADMM algorithm to efficiently solve a nonconvex optimization problem, based on readily measurable operating traces. Toward incremental real-world deployment, we take a non-intrusive approach to support flexible routing of micro-batches by adding a new DStream transformation we have developed to the existing Spark Streaming framework. As a result, our implementation can enforce scheduling decisions by modifying application workflows only. We have deployed our implementation on Amazon EC2 with emulated bandwidth constraints, and our experimental results on various types of queries have demonstrated the effectiveness of our proposed framework, as compared to the existing Spark Streaming scheduler and other data-locality-based heuristics.