BDS+: An Inter-Datacenter Data Replication System With Dynamic Bandwidth Separation
نویسندگان
چکیده
Many important cloud services require replicating massive data from one datacenter (DC) to multiple DCs. While the performance of pair-wise inter-DC transfers has been much improved, prior solutions are insufficient optimize bulk-data multicast, as they fail explore rich overlay paths that exist in geo-distributed DCs, well remaining bandwidth reserved for online traffic under fixed separation scheme. To take advantage these opportunities, we present BDS+, a near-optimal network system large-scale replication. BDS+ is an application-level multicast with fully centralized architecture, allowing central controller maintain up-to-date global view delivery status intermediate servers, order utilize available paths. Furthermore, each path, it leverages dynamic make use traffic. By constantly estimating demand and rescheduling accordingly, can further speed up multicast. Through pilot deployment largest service providers real-trace simulations, show achieve 3- 5× speedup over provider's existing several well-known routing baselines static separation. Moreover, reduce completion time bulk by 1.2 1.3 times.
منابع مشابه
Cost-minimizing Bandwidth Guarantee for Inter-datacenter Traffic
The emerging deployment of large-scale cloud applications incurs significant inter-datacenter traffic, which makes the scarce wide-area bandwidth across data centers become the performance bottleneck. To achieve the desirable network performance, bandwidth guarantee should be provided for the resulting inter-datacenter traffic. However, the existing bandwidth allocation methods mainly focus on ...
متن کاملInter-Datacenter Scheduling of Large Data Flows
Inter-datacenter transfers of non-interactive but timely large flows over a private (managed) network is an important problem faced by many cloud service providers. The considered flows are non-interactive because they do not explicitly target the end users. However, most of them must be performed on a timely basis and are associated with a deadline. We propose to schedule these flows by a cent...
متن کاملDynamic Replication based on Firefly Algorithm in Data Grid
In data grid, using reservation is accepted to provide scheduling and service quality. Users need to have an access to the stored data in geographical environment, which can be solved by using replication, and an action taken to reach certainty. As a result, users are directed toward the nearest version to access information. The most important point is to know in which sites and distributed sy...
متن کاملPricing Intra-Datacenter Networks with Over-Committed Bandwidth Guarantee
Current IaaS clouds provide performance guarantee on CPU and memory but no quantitative network performance for VM instances. Our measurements from three production IaaS clouds show that for the VMs with same CPU and memory, or similar pricing, the difference in bandwidth performance can be as much as 16×, which reveals a severe price-performance anomaly due to a lack of pricing for bandwidth g...
متن کاملClock-RSM: Low-Latency Inter-Datacenter State Machine Replication Using Loosely Synchronized Physical Clocks (Technical Report)
This paper proposes Clock-RSM, a new state machine replication protocol that uses loosely synchronized physical clocks to totally order commands for geo-replicated services. Clock-RSM assumes realistic non-uniform latencies among replicas located at different data centers. It provides low-latency linearizable replication by overlapping 1) logging a command at a majority of replicas, 2) determin...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
ژورنال
عنوان ژورنال: IEEE ACM Transactions on Networking
سال: 2021
ISSN: ['1063-6692', '1558-2566']
DOI: https://doi.org/10.1109/tnet.2021.3054924