Review: Active Network Vision & Reality

Active networks are an approach to network architecture that allows
customized programs to be executed within the network. The original
vision was to use a (1) capsule model of programmability to form a
network (2) accessible to all users that (3) enables a new range of
applications, which leverage computation within the network, thereby
accelerating the pace of innovation by separating services from the
underlying infrastructure. However, active networks raise serious
performance and security concerns because of the overhead of executing
untrusted code present within the network. The paper makes substantial
progress towards providing a more flexible network layer and also
addressing the performance and security concerns rasied by the presence
of mobile code in the network.
The authors aims to redress the situation by reporting what they have
learnt by designing, implementing and using the ANTS active network
toolkit. They claim ANTS is well-suited to this purpose because it is
based on an aggressive capsule design that adds extensibility at the IP
packet level.
The authors begin by giving some background on active networks and
describing their implementation of ANTS. The main contributions of the
paper follow:
1. A detailed discussion regarding the performance of capsules and their
suitability as a mechanism on which to build active networks.
2. A discussion on who can introduce new services. They talk about
ensuring that user code cannot harm the users of other services, and
that the consumption of shared resources should be reasonable enough to
not affect the performance of other services.
3. A discussion of what services can be introduced. The authors say that
any service should be expressible (use only a limited node API),
compact, fast and incrementally deployable.
The paper concludes that capsules can provide comparable forwarding
mechanisms when software-based routers are viable and that they are best
suited for experimentation. Furthermore, the authors were partially
successful in designing a network in which any untrsutedf users can
customize the processing of their packets within the network.

The paper was not that well-written. The overview in the introduction
was given before the reader is properly introduced to the concept of
active networks. Furthermore, they did not provide compelling arguments
to convince me that the idea of active networks was powerful enough to
overcome its flaws, which is probably also the reason they have not
taken off.
Received on Wed Nov 22 2006 - 14:49:11 EST