• Corpus ID: 13258854

Designing Distributed Systems Using Approximate Synchrony in Data Center Networks

@inproceedings{Ports2015DesigningDS,
  title={Designing Distributed Systems Using Approximate Synchrony in Data Center Networks},
  author={Dan R. K. Ports and Jialin Li and Vincent Liu and Naveen Kr. Sharma and Arvind Krishnamurthy},
  booktitle={NSDI},
  year={2015}
}
Distributed systems are traditionally designed independently from the underlying network, making worst-case assumptions (e.g., complete asynchrony) about its behavior. However, many of today's distributed applications are deployed in data centers, where the network is more reliable, predictable, and extensible. In these environments, it is possible to co-design distributed systems with their network layer, and doing so can offer substantial benefits. This paper explores network-level… 
Partitioned Paxos via the Network Data Plane
TLDR
The key insight behind Partitioned Paxos is to separate the two aspects of Paxos, agreement, and execution, and optimize them separately, and provide at least x3 latency improvement and x11 throughput improvement for a replicated instance of a RocksDB key-value store.
Just Say NO to Paxos Overhead: Replacing Consensus with Network Ordering
TLDR
A new replication protocol, Network-Ordered Paxos (NOPaxos), exploits network ordering to provide strongly consistent replication without coordination, providing replication without the performance cost in the data center.
1 Co-Designing State Machine Replication with the Datacenter Network : Speculative Paxos and NOPaxos
TLDR
This work analyzes the core protocols in distributed systems, and identifies key abstractions that can be efficiently implemented in network devices, given the limited hardware resources and restricted computational model, and can result in significant end-to-end performance improvements.
Series in Informatics Partitioned Paxos via the Network Data Plane
TLDR
Partitioned Paxos is a novel approach to network-accelerated consensus that separates the two aspects of Paxos, agreement and execution, and optimize them separately, and is able to provide at least ×3 latency improvement and ×11 throughput improvement for a replicated instance of a RocksDB keyvalue store.
On the many faces of atomic multicast
TLDR
GeoPaxos is introduced, a protocol that partitions the ordering of operations like atomic multicast while still keeping the state fully replicated and Kernel Paxos is presented, a Paxos implementation provided as a loadable kernel module, providing at the same time high performance, and abstracting ordering from the application execution.
Eris: Coordination-Free Consistent Transactions Using In-Network Concurrency Control
TLDR
Eris can process a large class of distributed transactions in a single round-trip from the client to the storage system without any explicit coordination between shards or replicas in the normal case, providing atomicity, consistency, and fault tolerance with less than 10% overhead.
Domino: using network measurements to reduce state machine replication latency in WANs
TLDR
This paper introduces Domino, a low-latency state machine replication protocol for wide-area networks that can achieve significantly lower commit latency than other consensus protocols, such as Mencius, Fast Paxos, and EPaxos.
Towards causal datacenter networks
TLDR
This work asks whether it is possible for the datacenter network to order all messages in a protocol-agnostic way, which would ensure causal delivery of all messages, making consistency a network-level guarantee.
1Pipe: scalable total order communication in data center networks
TLDR
Evaluation on a 32-server testbed shows that 1Pipe achieves scalable throughput and low latency in transactional key-value store, TPC-C, remote data structures, and replication that outperforms traditional designs by 2∼20x.
Network Hardware-Accelerated Consensus
TLDR
This paper posits that there are significant performance benefits to be gained by offering consensus as a network service (CAANS), which provides a complete Paxos protocol, is a drop-in replacement for software-based implementations of Paxos, makes no restrictions on network topologies, and is implemented in a higher-level, data-plane programming language.
...
1
2
3
4
5
...

References

SHOWING 1-10 OF 46 REFERENCES
PortLand: a scalable fault-tolerant layer 2 data center network fabric
TLDR
Through the design and implementation of PortLand, a scalable, fault tolerant layer 2 routing and forwarding protocol for data center environments, it is shown that PortLand holds promise for supporting a ``plug-and-play" large-scale, data center network.
VL2: a scalable and flexible data center network
TLDR
VL2 is a practical network architecture that scales to support huge data centers with uniform high capacity between servers, performance isolation between services, and Ethernet layer-2 semantics and can be deployed today, and a working prototype is built.
A scalable, commodity data center network architecture
TLDR
This paper shows how to leverage largely commodity Ethernet switches to support the full aggregate bandwidth of clusters consisting of tens of thousands of elements and argues that appropriately architected and interconnected commodity switches may deliver more performance at less cost than available from today's higher-end solutions.
F10: A Fault-Tolerant Engineered Network
TLDR
This work creates an engineered network and routing protocol that can almost instantaneously reestablish connectivity and load balance, even in the presence of multiple failures, and shows that following network link and switch failures, F10 has less than 1/7th the packet loss of current schemes.
Exploiting virtual synchrony in distributed systems
TLDR
It is argued that this approach to building distributed and fault-tolerant software is more straightforward, more flexible, and more likely to yield correct solutions than alternative approaches.
Tolerating Latency in Replicated State Machines Through Client Speculation
TLDR
This paper examines how to use speculative execution at the clients of a replicated service to reduce the impact of network and protocol latency, and gives design principles for using client speculation with replicated services, such as generating early replies and prioritizing throughput over latency.
ZooKeeper: Wait-free Coordination for Internet-scale Systems
TLDR
ZooKeeper provides a per client guarantee of FIFO execution of requests and linearizability for all requests that change the ZooKeeper state to enable the implementation of a high performance processing pipeline with read requests being satisfied by local servers.
Chasing the FLP impossibility result in a LAN: or, How robust can a fault tolerant server be?
TLDR
By trying to illustrate the practical impact of impossibility results, a very robust replicated service is discovered that it never stops delivering messages, not even under arbitrarily high load and very small timeout values.
It's Time for Low Latency
TLDR
This paper argues that OS researchers must lead the charge in rearchitecting systems to push the boundaries of low-latency datacenter communication and argues that 5-10µs remote procedure calls are possible in the short term - two orders of magnitude better than today.
Understanding network failures in data centers: measurement, analysis, and implications
TLDR
The first large-scale analysis of failures in a data center network is presented, finding that data center networks show high reliability, commodity switches such as ToRs and AggS are highly reliable, and network redundancy is only 40% effective in reducing the median impact of failure.
...
1
2
3
4
5
...