A Survey of Fast Recovery Mechanisms in the Data Plane

Chiesa, Marco; Kamisiński, Andrzej; Rak, Jacek; Rétvári, Gábor; Schmid, Stefan

doi:10.36227/techrxiv.12367508

“…Resilient routing has been widely studied [29], especially for fast recovery and reroute mechanisms [11]. We next focus on 1) local fast reroute (FRR) mechanisms, which covers statically pre-installed failover rules, and 2) non-local recovery mechanisms by means of (control/data plane) convergence.…”

Section: Related Workmentioning

confidence: 99%

“…However, unexpected failures (link/switch) are inevitable and happen regularly requiring a rapid action to ensure seamless connectivity without compromising on performance. A plethora of In-network Fast Reroute (FRR) approaches [11,29] have been developed entirely in the data plane to address such a problem. However, these approaches are slow, incur loops, trigger packet loss when routes become unavailable, to reroute traffic via a sub-optimal path [36] which may be shared by other traffic.…”

Section: Introductionmentioning

confidence: 99%

Shortcutting Fast Failover Routes in the Data Plane

Shukla

¹

,

Foerster

²

2021

Proceedings of the Symposium on Architectures for Networking and Communications Systems

View full text Add to dashboard Cite

In networks, availability is of paramount importance. As link failures are disruptive, modern networks in turn provide Fast ReRoute (FRR) mechanisms to rapidly restore connectivity. However, existing FRR approaches heavily impact performance until the slower convergence protocols kick in. The fast failover routes commonly involve unnecessary loops and detours, disturbing other traffic while causing costly packet loss. In this paper, we make a case for augmenting FRR mechanisms to avoid such inefficiencies. We introduce ShortCut that routes the packets in a loop free fashion, avoiding costly detours and decreasing link load. ShortCut achieves this by leveraging data plane programmability: when a loop is locally observed, it can be removed by short-cutting the respective route parts. As such, ShortCut is topology-independent and agnostic to the type of FRR currently deployed. Our first experimental simulations show that ShortCut can outperform control plane convergence mechanisms; moreover avoiding loops and keeping packet loss minimal opposed to existing FRR mechanisms. CCS CONCEPTS• Computer systems organization → Reliability.

show abstract

“…Resilient routing has been widely studied [30], especially for fast recovery and reroute mechanisms [11]. We next focus on 1) local fast reroute (FRR) mechanisms, which covers statically pre-installed failover rules, and 2) non-local recovery mechanisms by means of (control/data plane) convergence.…”

Section: Related Workmentioning

confidence: 99%

“…However, unexpected failures (link/switch) are inevitable and happen regularly requiring a rapid action to ensure seamless connectivity without compromising on performance. A plethora of In-network Fast Reroute (FRR) approaches [11,30] have been developed entirely in the data plane to address such a problem. However, these approaches are slow, incur loops, trigger packet loss when routes become unavailable, to reroute traffic via a sub-optimal path [37] which may be shared by other traffic.…”

Section: Introductionmentioning

confidence: 99%

Shortcutting Fast Failover Routes in the Data Plane

Shukla,

Foerster

2021

Preprint

0

View full text Add to dashboard Cite

In networks, availability is of paramount importance. As link failures are disruptive, modern networks in turn provide Fast ReRoute (FRR) mechanisms to rapidly restore connectivity. However, existing FRR approaches heavily impact performance until the slower convergence protocols kick in. The fast failover routes commonly involve unnecessary loops and detours, disturbing other traffic while causing costly packet loss. In this paper, we make a case for augmenting FRR mechanisms to avoid such inefficiencies. We introduce ShortCut that routes the packets in a loop free fashion, avoiding costly detours and decreasing link load. ShortCut achieves this by leveraging data plane programmability: when a loop is locally observed, it can be removed by short-cutting the respective route parts. As such, ShortCut is topology-independent and agnostic to the type of FRR currently deployed. Our first experimental simulations show that ShortCut can outperform control plane convergence mechanisms; moreover avoiding loops and keeping packet loss minimal opposed to existing FRR mechanisms. CCS CONCEPTS• Computer systems organization → Reliability.

show abstract

“…To meet their stringent availability requirements, most modern communication networks hence feature local fast rerouting algorithms in the data plane which typically operates at timescales several orders of magnitude shorter than the control plane [11], [12]: since rerouting decisions are local, failure recovery can in principle occur at the speed of packet forwarding. At the heart of such fast-reroute (FRR) mechanisms [13] lies the idea of pre-computing alternative paths at any node towards any destination. When a node locally detects a failed link or port, it can autonomously remove the corresponding entries from the forwarding table and continue using the remaining next hops for forwarding packets: a fast local reaction [14].…”

Section: Introductionmentioning

confidence: 99%

Grafting Arborescences for Extra Resilience of Fast Rerouting Schemes

Foerster¹,

Kamisiński²,

Pignolet³

et al. 2021

IEEE INFOCOM 2021 - IEEE Conference on Computer Communications

View full text Add to dashboard Cite

To provide a high availability and to be able to quickly react to link failures, most communication networks feature fast rerouting (FRR) mechanisms in the data plane. However, configuring these mechanisms to provide a high resilience against multiple failures is algorithmically challenging, as rerouting rules can only depend on local failure information and need to be predefined. This paper is motivated by the observation that the common approach to design fast rerouting algorithms, based on spanning trees and covering arborescences, comes at a cost of reduced resilience as it does not fully exploit the available links in heterogeneous topologies. We present several novel fast rerouting algorithms which are not limited by spanning trees, but rather extend and combine ("graft") multiple spanning arborescences to improve resilience. We compare our algorithms analytically and empirically, and show that they can significantly improve not only the resilience, but also accelerate the preprocessing to generate the local fast failover rules.1 Grafting is a technique to join two trees (and plants in general) into one. For example, many fruit trees today are grafted onto rootstock.

show abstract

A Survey of Fast Recovery Mechanisms in the Data Plane

Cited by 5 publications

References 149 publications

Shortcutting Fast Failover Routes in the Data Plane

Shortcutting Fast Failover Routes in the Data Plane

Shortcutting Fast Failover Routes in the Data Plane

Grafting Arborescences for Extra Resilience of Fast Rerouting Schemes

Contact Info

Product

Resources

About