ebook img

Duet: Cloud Scale Load Balancing with Hardware PDF

25 Pages·2014·0.6 MB·English
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview Duet: Cloud Scale Load Balancing with Hardware

Duet: Cloud Scale Load Balancing with Hardware and Software Rohan Gandhi Hongqiang Harry Liu Guohan Lu J i t u P a d h ye Y. Charlie Hu Lihua Yuan Ming Zhang 1 Load Balancer is Critical For Online Services Terminology Internet VIP – Virtual IP DIP – Direct IP Mux – Multiplexer VIP1, VIP2 Datacenter Store VIP-DIP mapping Load balancer (Mux) Split VIP traffic to DIPs Server Server Server Server . . . . . . DIP = 10.0.1.1 DIP = 10.0.2.2 DIP = 10.0.3.3 DIP = 10.0.4.4 bing.com OneDrive VIP1 = 1.1.1.1 VIP2 = 1.1.1.2 Load balancer provides high availability and scalability 2 Existing LBs Have Limitations Specialized Hardware LBs Too costly Poor robustness $100+ million for 15 Tbps 1+1 redundancy Software LBs Scale with demand High robustness Scale up/down according to n+1 redundancy VIP traffic Software LBs have cost and performance limitations 3 Software LB Design DC Network DIP2 Client SMux DIP1 VIP1 DIP1, DIP2 VIP1 . . . . Software LB Benefits: • High robustness • Scale with demand 4 Ananta: Cloud-scale load balancing [SIGCOMM’13] Software LB has Limitations 100 n 10 n o 80 o i t i alf )c taz 60 n e i i y sm 1 litU 40 c n ( U e t P 20 a C L 0.1 0 100k 300k 450k 100k 300k 450k Traffic (pkts/sec) Traffic (pkts/sec) High latency inflation: 200 usec Low capacity: 300k pkts/sec 5k SMuxes needed at 15Tbps traffic in 50k server DC 5 How can we build high performance, low cost and robust load balancer? Duet ideas • Use commodity switches as hardware Muxes • Use software Muxes as a backstop 6 Can Switch Act As a Mux Switches offer: • High capacity (500+ million pkts/sec) • Low latency inflation (1 usec) Mux functionalities Switch resources • Split VIP traffic across DIPs • ECMP • Forward VIP traffic to DIPs • Tunneling 7 Implementing HMux on Switch VIP DIPs Forwarding Table ECMP Table Tunneling Table 10.0.0.1 Index DIPs VIP 1.1.1.1/32 10.0.0.2 1.1.1.1/32 0 10.0.0.1 20.0.0.1 1.1.1.2/32 1 10.0.0.2 1.1.1.2/32 20.0.0.2 2 20.0.0.1 3 20.0.0.2 HMux Dest: Src: Dest: Src: Dest: Src: VIP Client DIP HMux VIP Client 8 Key Design Challenges • Limited switch memory • High failure robustness • VIP assignment • VIP migration 9 Challenge 1: Switches have Limited Memory Workload: 100k+ VIPs and 1+ millions DIPs Single HMux cannot store all VIPs and DIPs Forwarding Table ECMP Table Tunneling Table Index DIPs VIP 1.1.1.1/32 0 10.0.0.1 1.1.1.2/32 1 10.0.0.2 2 20.0.0.1 3 20.0.0.2 Limits #VIPs Limits #DIPs Table Forwarding ECMP Tunneling Max. size 16k 4k 512 Max VIPs Max DIPs 10

Description:
Y. Charlie Hu Load balancer provides high availability and scalability . Server . Server Ananta: Cloud-scale load balancing [SIGCOMM'13].
See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.