Citation Context

... data plane of network boxes and checks for correctness against invariants. Anteater [10] uses boolean expressions and SAT solvers for network modeling and checking. However, offline checking cannot prevent bugs from damaging the network until the periodic check runs. Online monitoring: Several tools help troubleshoot network programs at run-time. OFRewind [14] captures and reproduces the sequence of problematic OpenFlow command sequence. ATPG [16] systematically generates test packets against router configurations, and monitors network health by perioidically sending these tests packets. NDB [4] is a network debugger that sets breakpoints and packet backtraces in SDN, so that the abnormal behavior can be analyzed step-by-step. These tools complement but not replace the need for real-time policy verification. VeriFlow [9], is the work most closely related to NetPlumber. VeriFlow also verifies the compliance of network updates with specified policies in real time. It uses a trie structure to search rules based on equivalent classes (ECs), and upon an update, finds out the affected ECs and update the forwarding graph for that class. This in turn triggers a rechecking of affected policie...

Citation Context

...ds and replay (filtered) network events using a hypervisor. However,OFRewind is not concerned with recreating state in a correct manner and only provides coarse grained filtering ability. Thendb tool =-=[6]-=- provides a trace view of OpenFlow forwarding tables traversed by data packets in the network. Unlike HOTSWAP, ndb aims at troubleshooting the data plane, and not at reconstructing the state in the co...

Citation Context

...nnot provide regular traffic forwarding functionality. There has been also some work in the SDN realm related to policy management. However, most of these works focus on either static [20] or dynamic =-=[16, 19, 21]-=- checking for policy rule installation in network devices. To the best of our knowledge this is the first work that explores autonomic QoS policy enforcement framework in the realm of SDN. IV. PROPOSE...

Citation Context

...t there is a wide range of misconfigurations observed on middleboxes. As a result, a key requirement for network operations is to automate management and debugging of network problems e.g., using SDN =-=[37, 38]-=-. To improve network debugging, one approach is to do static and runtime verification of desired properties e.g., to check configurations, verify reachability, detect routing loops, guarantee isolatio...

Citation Context

...with this task, but existing techniques tend to be protocol-specific and cannot necessarily be applied to arbitrary distributed applications, or SDNs with arbitrary control programs. Hence, as others =-=[8,9]-=- have observed, a more powerful debugger is needed. Existing solutions, such as NetSight [9] or SNP [35], approach this problem by offering a kind of “backtrace”, analogous to a stack trace in a conve...

Citation Context

...li bilit Multiple Paths Network Provenance, Detection & Debugging Fig. 3. A NEBULA perspective on the network from Figure 2. 1 network provenance [42], failure detection [18,41] and network debugging =-=[20]-=-, problems can be rapidly diagnosed and repaired, even while the services continue operating. 3 NEBULA Future Internet Architecture Integration The first phase of NEBULA required invention of new appr...

Citation Context

...nd where data collection and data aggregation happens, VND is designed to scale to many tenants. VND is a significant improvement over existing proposals for enterprise network diagnosis, such as NDB =-=[12]-=-, OFRewind [21], Anteater [18] and HSA [13], which expose all the raw network information. This leads to obvious scale issues, but it also weakens isolation across tenants and exposes crucial informat...