A safe, efficient update protocol for openflow networks

Abstract

We describe a new protocol for update of OpenFlow networks, which has the packet consistency condition of [14] and a weak form of the flow consistency condition of [14]. The protocol conserves switch resources, particularly TCAM space, by ensuring that only a single set of rules is present on a switch at any time. The protocol exploits the identity of switch rules with Boolean functions, and the ability of any switch to send packets to a controller for routing. When a network changes from one ruleset (ruleset 1) to another (ruleset 2), the packets affected by the change are computed, and are sent to the controller. When all switches have been updated to send affected packets to the controller, ruleset 2 is sent to the switches and packets sent to the controller are re-released into the network.

A correct, zero-overhead protocol for network updates

“…Request permissions from permissions@acm.org. tion of a combinational logic network [5,9], a problem known to be in co-N P. This result is similar to [3], and has been implicitly exploited for network update [4,8,7].…”

“…All packets in the same flow must be handled by a single network configuration (Strong Consistency), or the prefix of a flow must be handled by one configuration and the suffix of the flow must be handled by the successor configuration (Weak Consistency, introduced in [4]) [8] derived an elegant method for ensuring both packet consistency and flow consistency. Both rulesets were loaded into each switch, with an unused bit in the header field used to indicate which ruleset was desired.…”

“…Effectively, the method of [8] cuts the available TCAM in half. [4] sought to remedy this by loading an intermediate ruleset onto the switches. In this method, packets whose handling was modified by the change were sent to the controller (or other packet refuge) by the intermediate ruleset; once the intermediate ruleset had been completely loaded onto all switches, the final ruleset was sent to the switches and the held packets were sent to their destinations (or any switch along their final path).…”

Practical, Real-time Centralized Control for CDN-based Live Video Delivery

“…VDN keeps previous versions of the FIB ("shadow FIBs") to allow clusters to forward requests with old version numbers (e.g., during global state transitions), similar to previous work [20,28,34]. When an unknown version is encountered, VDN resorts to using distributed control.…”

Abstractions for software-defined networks

“…These optimized mechanisms generate fewer messages, use less rule space on switches, or complete the transition more rapidly than full two-phase update. Consistent updates can also be implemented incrementally [22] or by diverting some packets to the controller [31].…”