Unleashing the Potential of In-Network Computing [Invited]

Speaker: Daehyeok Kim, Microsoft

Recent advances in programmable networking hardware create a new computing paradigm called in-network computing. This new paradigm allows functionality that has been served by commodity servers, ranging from network middleboxes to components of distributed systems, to be performed in the network. I argue that to fully unleash its potential, we need resource elasticity and fault resiliency via higher-level abstractions.

In this talk, I demonstrate that in-network computing can be elastic and resilient by designing high-level abstractions and runtime systems that enable us to effectively leverage compute and memory resources available outside of a single type of device -- e.g., programmable switches -- while hiding the complexities of dealing with device heterogeneity. I begin by introducing TEA, a framework that provides elastic memory by enabling memory-intensive in-switch applications, such as cloud-scale load balancers, to leverage DRAM on remote servers via virtual table abstraction. Then I present ExoPlane and RedPlane, frameworks that support evolving in-network computing workloads and requirements -- i.e., serving multiple concurrent applications and making them fault-tolerant -- via infinite switch resource and one big fault-tolerant switch abstractions. Several systems in the industry are now adopting some of the technologies presented in this talk.

Speaker Bio: Daehyeok Kim is a senior researcher at Microsoft. He recently completed his Ph.D. in the computer science department at Carnegie Mellon University. His research interests lie in the intersection of computer systems and networking with a focus on building new abstractions and runtime systems for in-network computing. He is a recipient of the Microsoft Research Ph.D. Fellowship.

----------------------------------------------------------

----------------------------------------------------------

The Intel/VMware Crossroads 3D-FPGA Academic Research Center is jointly supported by Intel and VMware. The center is committed to public and free dissemination of its research outcome.

----------------------------------------------------------
Chapters

0:00 Introduction
0:32 Traditional view of networks
1:39 New paradigm: In-network computing
2:19 Why is in-network computing promising?
3:40 Illustrative example: In-network DDoS defense
4:49 Ideal view of in-network computing
5:12 Today's constrained view of in-network computing
5:52 Analogy: Operating / Distributed systems for server-based com Great abstractions for resource elasticity, multiplexing, and fault resilience!
6:48 What makes it challenging for in-network computing?
7:40 Opportunity. Resources outside a single type of device
8:27 Research vision
9:59 Context: Network functions on programmable switches
11:15 Our work: Table Extension Architecture (TEA)
12:02 TEA design overview
12:49 Enabling external DRAM access from switch
13:57 Enabling RDMA in the switch data plane Key idea: Crafting RDMA packets using ASIC's programmability
15:58 Enabling single round-trip table lookup
16:41 Potential approach: Cuckoo hashing
17:36 TEA-table: lookup data structure
18:33 Challenge 3: Delayed packet processing
19:27 Offloading packet store to TEA-table for delayed pachet process Key idea: Employing scratchpad in TEA-table to buffer packets
20:12 Scaling TEA with multiple servers
20:55 Putting it all together TEA provides a virtual table abstraction to memory-intensive applications
21:44 Implementation and evaluation setup
22:11 How does TEA affect packet processing latency?
24:13 Impacts & on-going work
25:53 Problem: Enabling "multiple apps" on a switch Example scenario in a data center
26:57 ExoPlane: OS for switch resource augmentation architect
28:46 Problem: Switch failure
29:52 RedPlane: Making in-network applications fault tolerant
30:41 Network as an elastic and resilient computing platform
30:51 Case study: In-network DDoS defense
32:09 Reliable in-network computing at scale
33:35 Runtime reconfigurable data-plane architecture

[cmporeddy]

Where can I download this presentation PPT?