Cloud data centers host diverse applications, mixing workloads that require small predictable latency with others requiring large sustained throughput. In this environment, today's state-of-the-art TCP protocol falls short. We present measurements of a 6000 server production cluster and reveal impairments that lead to high application latencies, rooted in TCP's demands on the limited buffer space available in data center switches. For example, bandwidth hungry "background" flows build up queues at the switches, and thus impact the performance of latency sensitive "foreground" traffic.To address these problems, we propose DCTCP, a TCP-like protocol for data center networks. DCTCP leverages Explicit Congestion Notification (ECN) in the network to provide multi-bit feedback to the end hosts. We evaluate DCTCP at 1 and 10Gbps speeds using commodity, shallow buffered switches. We find DCTCP delivers the same or better throughput than TCP, while using 90% less buffer space. Unlike TCP, DCTCP also provides high burst tolerance and low latency for short flows. In handling workloads derived from operational measurements, we found DCTCP enables the applications to handle 10X the current background traffic, without impacting foreground traffic. Further, a 10X increase in foreground traffic does not cause any timeouts, thus largely eliminating incast problems.

/pdf/data-center-tcp-dctcp-4d3pj8xdal.pdf

Data center TCP (DCTCP)

At some point in the future, how far out we do not exactly know, wireless access to the Internet will outstrip all other forms of access bringing the freedom of mobility to the way we access the we...

ACM SIGCOMM computer communication review

An important class of datacenter applications, called Online Data-Intensive (OLDI) applications, includes Web search, online retail, and advertisement. To achieve good user experience, OLDI applications operate under soft-real-time constraints (e.g., 300 ms latency) which imply deadlines for network communication within the applications. Further, OLDI applications typically employ tree-based algorithms which, in the common case, result in bursts of children-to-parent traffic with tight deadlines. Recent work on datacenter network protocols is either deadline-agnostic (DCTCP) or is deadline-aware (D3) but suffers under bursts due to race conditions. Further, D3 has the practical drawbacks of requiring changes to the switch hardware and not being able to coexist with legacy TCP. We propose Deadline-Aware Datacenter TCP (D2TCP), a novel transport protocol, which handles bursts, is deadline-aware, and is readily deployable. In designing D2TCP, we make two contributions: (1) D2TCP uses a distributed and reactive approach for bandwidth allocation which fundamentally enables D2TCP's properties. (2) D2TCP employs a novel congestion avoidance algorithm, which uses ECN feedback and deadlines to modulate the congestion window via a gamma-correction function. Using a small-scale implementation and at-scale simulations, we show that D2TCP reduces the fraction of missed deadlines compared to DCTCP and D3 by 75% and 50%, respectively.

/pdf/deadline-aware-datacenter-tcp-d2tcp-42ztsu9nzb.pdf

Deadline-aware datacenter tcp (D2TCP)

Resource management in a cloud environment is a hard problem, due to: the scale of modern data centers; the heterogeneity of resource types and their interdependencies; the variability and unpredictability of the load; as well as the range of objectives of the different actors in a cloud ecosystem. Consequently, both academia and industry began significant research efforts in this area. In this paper, we survey the recent literature, covering 250+ publications, and highlighting key results. We outline a conceptual framework for cloud resource management and use it to structure the state-of-the-art review. Based on our analysis, we identify five challenges for future investigation. These relate to: providing predictable performance for cloud-hosted applications; achieving global manageability for cloud systems; engineering scalable resource management systems; understanding economic behavior and cloud pricing; and developing solutions for the mobile cloud paradigm .

Resource Management in Clouds: Survey and Research Challenges

This paper describes a new approach to end-to-end congestion control on a multi-user network. Rather than manually formulate each endpoint's reaction to congestion signals, as in traditional protocols, we developed a program called Remy that generates congestion-control algorithms to run at the endpoints.In this approach, the protocol designer specifies their prior knowledge or assumptions about the network and an objective that the algorithm will try to achieve, e.g., high throughput and low queueing delay. Remy then produces a distributed algorithm---the control rules for the independent endpoints---that tries to achieve this objective.In simulations with ns-2, Remy-generated algorithms outperformed human-designed end-to-end techniques, including TCP Cubic, Compound, and Vegas. In many cases, Remy's algorithms also outperformed methods that require intrusive in-network changes, including XCP and Cubic-over-sfqCoDel (stochastic fair queueing with CoDel for active queue management). Remy can generate algorithms both for networks where some parameters are known tightly a priori, e.g. datacenters, and for networks where prior knowledge is less precise, such as cellular networks. We characterize the sensitivity of the resulting performance to the specificity of the prior knowledge, and the consequences when real-world conditions contradict the assumptions supplied at design-time.

/pdf/tcp-ex-machina-computer-generated-congestion-control-1kiokvnql8.pdf

TCP ex machina: computer-generated congestion control

Achieving efficient and fair bandwidth allocation while minimizing packet loss and bottleneck queue in high bandwidth-delay product networks has long been a daunting challenge. Existing end-to-end congestion control (e.g., TCP) and traditional congestion notification schemes (e.g., TCP+AQM/ECN) have significant limitations in achieving this goal. While the XCP protocol addresses this challenge, it requires multiple bits to encode the congestion-related information exchanged between routers and end-hosts. Unfortunately, there is no space in the IP header for these bits, and solving this problem involves a non-trivial and time-consuming standardization process. In this paper, we design and implement a simple, low-complexity protocol, called variable-structure congestion control protocol (VCP), that leverages only the existing two ECN bits for network congestion feedback, and yet achieves comparable performance to XCP, i.e., high utilization, negligible packet loss rate, low persistent queue length, and reasonable fairness. On the downside, VCP converges significantly slower to a fair allocation than XCP. We evaluate the performance of VCP using extensive ns2 simulations over a wide range of network scenarios and find that it significantly outperforms many recently-proposed TCP variants, such as HSTCP, FAST, CUBIC, etc. To gain insight into the behavior of VCP, we analyze a simplified fluid model and prove its global stability for the case of a single bottleneck shared by synchronous flows with identical round-trip times.

One more bit is enough

Achieving efficient and fair bandwidth allocation while minimizing packet loss in high bandwidth-delay product networks has long been a daunting challenge. Existing end-to-end congestion control (eg TCP) and traditional congestion notification schemes (eg TCP+AQM/ECN) have significant limitations in achieving this goal. While the recently proposed XCP protocol addresses this challenge, XCP requires multiple bits to encode the congestion-related information exchanged between routers and end-hosts. Unfortunately, there is no space in the IP header for these bits, and solving this problem involves a non-trivial and time-consuming standardization process.In this paper, we design and implement a simple, low-complexity protocol, called Variable-structure congestion Control Protocol (VCP), that leverages only the existing two ECN bits for network congestion feedback, and yet achieves comparable performance to XCP, ie high utilization, low persistent queue length, negligible packet loss rate, and reasonable fairness. On the downside, VCP converges significantly slower to a fair allocation than XCP. We evaluate the performance of VCP using extensive ns2 simulations over a wide range of network scenarios. To gain insight into the behavior of VCP, we analyze a simple fluid model, and prove a global stability result for the case of a single bottleneck link shared by flows with identical round-trip times.

/pdf/one-more-bit-is-enough-1dbi09jfb3.pdf

Denial of Service (DoS) attacks frequently happen on the Internet, paralyzing Internet services and causing millions of dollars of financial loss. This work presents NetFence, a scalable DoS-resistant network architecture. NetFence uses a novel mechanism, secure congestion policing feedback, to enable robust congestion policing inside the network. Bottleneck routers update the feedback in packet headers to signal congestion, and access routers use it to police senders' traffic. Targeted DoS victims can use the secure congestion policing feedback as capability tokens to suppress unwanted traffic. When compromised senders and receivers organize into pairs to congest a network link, NetFence provably guarantees a legitimate sender its fair share of network resources without keeping per-host state at the congested link. We use a Linux implementation, ns-2 simulations, and theoretical analysis to show that NetFence is an effective and scalable DoS solution: it reduces the amount of state maintained by a congested router from per-host to at most per-(Autonomous System).

/pdf/netfence-preventing-internet-denial-of-service-from-inside-3uteo8eg9e.pdf

NetFence: preventing internet denial of service from inside out

Multicast benefits data center group communications in saving network bandwidth and increasing application throughput. However, it is challenging to scale Multicast to support tens of thousands of concurrent group communications due to limited forwarding table memory space in the switches, particularly the low-end ones commonly used in modern data centers. Bloom Filter is an efficient tool to compress the Multicast forwarding table, but significant traffic leakage may occur when group membership testing is false positive. To reduce the Multicast traffic leakage, in this paper we bring forward a novel multi-class Bloom Filter (MBF), which extends the standard Bloom Filter by embracing element uncertainty. Specifically, MBF sets the number of hash functions in a per-element level, based on the probability for each Multicast group to be inserted into the Bloom Filter. We design a simple yet effective algorithm to calculate the number of hash functions for each Multicast group. We have prototyped a software based MBF forwarding engine on the Linux platform. Simulation and prototype evaluation results demonstrate that MBF can significantly reduce Multicast traffic leakage compared to the standard Bloom Filter, while causing little system overhead.

/pdf/scalable-data-center-multicast-using-multi-class-bloom-5aevb1y8ch.pdf

Scalable data center multicast using multi-class Bloom Filter

Optical data center networks (DCNs) are becoming increasingly attractive due to their technological strengths compared to traditional electrical networks However, prior optical DCNs are either hard to scale, vulnerable to single point of failure, or provide limited network bisection bandwidth for many practical DCN workloads To this end, we present WaveCube, a scalable, fault-tolerant, high-performance optical DCN architecture To scale, WaveCube removes MEMS1, a potential bottleneck, from its design Wave-Cube is fault-tolerant since it does not have single point of failure and there are multiple node-disjoint parallel paths between any pair of Top-of-Rack (ToR) switches WaveCube delivers high performance by exploiting multi-pathing and dynamic link bandwidth along the path Our extensive evaluation results show that WaveCube outperforms previous optical DCNs by up to 400% and delivers network bisection bandwidth that is 70%–85% of an ideal non-blocking network under both realistic and synthetic traffic patterns WaveCube's performance degrades gracefully under failures — it drops 20% even with 20% links cut WaveCube also holds promise in practice — its wiring complexity is orders of magnitude lower than Fattree, BCube and c-Through at large scale, and its power consumption is 35% of them

/pdf/wavecube-a-scalable-fault-tolerant-high-performance-optical-pkjnyb0vlg.pdf

Yong Xia

Papers

One more bit is enough

One more bit is enough

NetFence: preventing internet denial of service from inside out

Scalable data center multicast using multi-class Bloom Filter

WaveCube: A scalable, fault-tolerant, high-performance optical data center architecture