Publications

FAVE: Bandwidth-Aware Failover in Virtualized SDN for Clouds

2019 IEEE 12th International Conference on Cloud Computing (CLOUD), work-in-progress, Milan, Italy, 2019, pp. 505-507, doi: 10.1109/CLOUD.2019.00092.

Heesang Jin, Gyeongsik Yang, Bong-yeol Yu, Chuck Yoo

Download PDF

Abstract

Network virtualization based on SDN has gained attention in cloud networking. However, existing studies have not provided any failover technique in the event of physical link failure. We propose FAVE, which provides seamless failover and bandwidth-aware protection. FAVE carefully allocates backup routes to handle both failure and interference between tenants. Evaluation shows that FAVE is effective. To our knowledge, FAVE is the first attempt to address failover in virtualized SDN environments.

Paper summary

Reserve a backup route that can carry the traffic

A failover route must preserve bandwidth as well as connectivity.

FAVE addresses failures in shared SDN virtual networks. Moving traffic to a backup path can interfere with another tenant, so the backup calculation considers both tenant bandwidth requirements and existing reservations.

Reserve capacity for both primary and backup routes

The route calculator excludes links on the tenant’s primary route and links without sufficient available capacity. It also checks whether existing backup routes would interfere with a newly requested primary route and recalculates them when necessary.

Conditions used when FAVE selects or revises a backup route.
CheckAction
Primary-route overlapExclude primary links
Available capacityExclude links with insufficient bandwidth
Conflict with another primary routeRecalculate the affected backup

Keep both tenants forwarding after a link failure

The OpenVirteX implementation uses ONOS tenant controllers in a fat-tree test network. A link serving tenant 1 is broken during the experiment. Its reserved backup route takes over, and both tenants retain nearly the same throughput in the reported test. FAVE also moves a backup reservation when it would interfere with another tenant’s primary route.

Throughput of two tenants during a test with a link failure at 30 seconds.View full-size figure
Tenant 1’s link fails at 30 seconds. Both tenants maintain approximately their previous throughput as FAVE switches to the reserved backup route.

Key takeaways

  1. A backup route needs capacity as well as connectivity. FAVE includes reserved bandwidth in failover planning.
  2. Recovery for one tenant must account for other tenants’ routes. Coordinating primary and backup reservations supports the nearly unchanged throughput observed during the tested link failure.