In-network Congestion-aware Load Balancing at Transport Layer
share
Load balancing at transport layer is an important function in data centers, content delivery networks, and mobile networks. This function has unique requirements and greatly impacts the end users' experience. Recent studies in the field have identified per-connection consistency (PCC) as the main requirement to produce efficient L4 load balancing and have proposed various solutions to achieve that goal. This paper identifies load imbalance among service instances as the main cause of additional processing times. Furthermore, existing load balancers rely on one of two methods: host-level traffic redirection, which may add as much as 8.5 underlying networks, or connection tracking, which consumes a considerable amount of memory in load balancers and is prone to denial of service attacks. Both of these methods result in inefficient usage of networking resources. We propose the In-Network Congestion-Aware load Balancer (INCAB) to achieve even load distribution and optimize network resources used for load balancing in addition to meeting the PCC requirement. We show that our congestion-aware solution capable of identifying and monitoring each instance's most-utilized resource improves the load distribution among all service instances. Our solution utilizes a Bloom filter and an ultra-compact connection table for in-network flow distribution and does not rely on end hosts for traffic redirection. Our flow level simulations show that INCAB improves flows' average completion time by 9.5
READ FULL TEXT