Presentation on theme: "Virtualization Group FIND Meeting. Does Virtualization aid Security? Depends what you build on top –Were only providing mechanisms, not solutions Better:"— Presentation transcript:
Does Virtualization aid Security? Depends what you build on top –Were only providing mechanisms, not solutions Better: –Some level of isolation –Control set of users –Better defaults, reduce exposed vulnerabilities –Baked-in forensic capability; liability determination Worse: –Lots of vulnerabilities in optics –Unclear what the TCB is---finger pointing –Amplifies attack power based on hierarchy –Complex auditing task (many layers to demux) Tomography; mapping topologies frustrating Upgrading versions across multiple partitions
Do Sensors need Virtualization? Just build metanetwork specialized to sensors Useful for wireless sensor networks? –Same arguments apply as regular IP –Yes for GENI; not operationally? Interface between wireless/wired –Rich specification for optical network setup –Depends on how fast the setup/teardown needs –May hide changes in physical topology--e.g., delay
Suitability for Tomorrow Increasingly programmable components –Virtualization is a nice way to harness –Need to share? Enough to hand out individually? –Hard to envision technology cost being a constraint; increasingly less important Speeds surpass ability to source/sink –Need to divide; different interfaces (SAN, etc.) Isolation –Is it an economic issue or a technical one –Horizontal vs. vertical markets –Cloud-based services; distributed management
Active Networks? –What does it mean to do things in the network –Does topology matter? It does for data synthesis Maybe for location based services But BW is free, so we can tunnel things lots of places pretty darn easily –Is a middlebox core or edge? –Active network struggle all over again?