Virtual Routers on the Move: Live Router Migration as a Network-Management Primitive
by Yi Wang, Eric Keller, Brian Biskeborn, Jacibus van, Jennifer Rexford
url show details
Details
| type: | misc | | abstract: | The complexity of network management is widely recognized
as one of the biggest challenges facing the Internet today.
Point solutions for individual problems further increase system
complexity while not addressing the underlying causes.
In this paper, we argue that many network-management
problems stem from the same root cause—the need to maintain
consistency between the physical and logical configuration
of the routers. Hence, we propose VROOM (Virtual
ROuters On the Move), a new network-management primitive
that avoids unnecessary changes to the logical topology
by allowing (virtual) routers to freely move from one physical
node to another. In addition to simplifying existing
network-management tasks like planned maintenance and
service deployment, VROOM can also help tackle emerging
challenges such as reducing energy consumption. We present
the design, implementation, and evaluation of novel migration
techniques for virtual routers with either hardware or
software data planes. Our evaluation shows that VROOM
is transparent to routing protocols and results in no performance
impact on the data traffic when a hardware-based
data plane is used. | | url: | http://www.cs.princeton.edu/~yiwang/papers/vroom_sigcomm08.pdf |
|
|
You need to log in to add tags and post comments.
The paper proposes the VROOM architecture which allows live migration of logical routers across different physical routers. To do so, the authors break the tight coupling between logical and physical routers, and instantiate the logical router as a virtual router on top of a dataplane hypervisor on the physical router.
The paper is quite intersting in that the objective is to allow migrating this logical router, without traffic disruption, to different physical routers which need not be identical in hardware. The authors note that for ensuring freedom from traffic disruption, the dataplane functionality should not be interrupted; however, short duration control plane disruptions are tolerable, and are rectified by retransmission mechanisms available in routing protocols.
The paper then describes the Vroom architecture, and the migration process which mainly involves control plane migration, data plane cloning, and link migration. The use of switching to hotswap links is nice, although not novel. The separation of control and data plane, as the authors mention, is quite similar to what is implemented in routers today with a route processor card which forms the control plane and one or more line cards with their own cpu forming the dataplane.
I do agree with lorenzo's comment here that virtualization is not a strict requirement here. At times it maps too closely to the logical router to make it appear confusing. But using virtualization makes the whole process of migration simpler.
I found the paper a bit dragging towards the evaluation side. Some data presented seemed rather redundant. For eg. once the control plane migration time is known, the ospf timeout that the scheme can tolerate is an emergent property.
But overall, the idea presented is quite interesting, eminently practical and rife with benefits. The mention of separating control plane state from control plane software in the conclusion is also quite thought provoking. Such an optimization to routers can ease software update, plus it will integrate well with vroom in that migration time can be further reduced.