1: \begin{abstract}

2: %

3: The recent line of research into topology design focuses on lowering network

4: diameter. Many low-diameter topologies such as Slim Fly or Jellyfish that

5: substantially reduce cost, power consumption, and latency have been proposed. A

6: key challenge in realizing the benefits of these topologies is \emph{routing}.

7: On one hand, these networks provide shorter path lengths than established

8: topologies such as Clos or torus, leading to performance improvements. On the

9: other hand, the number of shortest paths between each pair of endpoints is much

10: smaller than in Clos, but there is a large number of non-minimal paths between

11: router pairs. This hampers or even makes it impossible to use established

12: multipath routing schemes such as ECMP.

13: %

14: In this work, to facilitate high-performance routing in modern networks, we

15: analyze existing routing protocols and architectures, focusing on how well they

16: exploit the diversity of minimal and non-minimal paths. We first develop a

17: taxonomy of different forms of support for multipathing and overall path

18: diversity. Then, we analyze how existing routing schemes support this

19: diversity. Among others, we consider multipathing with both shortest and

20: non-shortest paths, support for disjoint paths, or enabling adaptivity. To address

21: the ongoing convergence of HPC and ``Big Data'' domains, we consider routing

22: protocols developed for both HPC systems and

23: for data centers as well as general clusters. Thus, we cover architectures and

24:   protocols based on Ethernet,

25: InfiniBand, and other HPC networks such as Myrinet. 

26: %

27: Our review will foster developing future high-performance multipathing routing

28: protocols in supercomputers and data centers.

29: %

30: \end{abstract}

31: