take advantage of location opportunities as they arise.

Also, have a strategy for how you can still obtain a top-level view from all the independent edge-based instances. This is especially critical in areas such as security, where you need a global view of a multiple-point attack. If there’s an attack that consists of multiple events that are all below individual thresholds, then there’s still some correlation required up top to be able to recognize it as an attack. You cannot get away with saying that these are distributed, independent problems at the edge and that no correlation is required.

neaSe: You will never remove the concept of location from networking. It will always be part and parcel of the value proposition. Bandwidth consumption will be rarer the farther you span, and latency will be shorter the closer you are. Physical location of networking resources never completely goes away. The network is never location independent and always has a component of geography, location, and physics. You cannot separate them.

cReeGeR: The issues involved in network virtualization are moving quickly. Mass interest in virtualizing the data center is breaking a lot of traditional physical versus logical bounds. You need to concentrate on what you’re trying to achieve in your data center and what key properties you’re trying to preserve. If you do decide to virtualize, do an internal cloud, or subcontract out to an external cloud vendor, then you need to parallel your architecture closely to industry leaders such as Amazon so you keep close to current accepted practices. Additionally, to avoid breakage between physical and virtual devices, you want to minimize functionality and performance investments that require device-specific configuration of the physical infrastructure. As virtual devices become more prevalent, those device-specific configurations will become more of a burden.

Some network vendors are offering products under the banner of network virtualization that provide virtual implementations of their physical devices. I believe they are being offered to preserve a customer’s investments in legacy infrastructure. By preserving the status quo, however, it will be that much more difficult to take advantage

maRTin caSaDo
if people want the
dynamics and cost
structure of the
cloud, they should
either not invest in
anything now; or
invest in a scale-
out commodity
and make it look
like amazon. if
they do not take
one of these two
paths, then they
will be locked into a
vertically integrated
stack and the world
will pass them by.

of new, more efficient and functional
architectures as they gain broader ac-
ceptance. The advice here suggests that
you keep things simple. Avoid invest-
ing in vendor-proprietary functions,
and wait to see what new architectures
emerge. Once you identify these new
architectures, invest conservatively as
they gain acceptance.
cRoSBy: The key point is to be aware
that your networking competence in-
vestments are going to shift radically.
The new network will be automated,
aware of the current locus of the work-
load, and dynamically reconfigure the
infrastructure as the workload mi-
grates or scales elastically.
TaVakoLi: An opportunity exists to
implement very fine-grained, high-
quality enforcement at the very edge
of the network, including on the host.
That will have to be stitched into your
service model. You can scale and dis-
tribute your control to the very edge
of the network, which now is on the
hosts. The question is, who ends up
driving the overall policy decision?

neaSe: If you’re a network person and you’re not touching VMware and find yourself not needed, you have to ask yourself whether or not your skill set is not needed as well. The network edge has moved, and if you are not architecting the network inside the server, then your skill set may not matter.

BeeLeR: The good news is that some systems administrators don’t have a clue about networking. This is an opportunity for network engineers still to add value in the new virtualized world.

 

Related articles on queue.acm.org

network Virtualization: Breaking the Performance Barrier

Scott Rixner

http://queue.acm.org/detail.cfm?id=1348592

CTO Roundtable: Virtualization

Mache Creeger

http://queue.acm.org/detail.cfm?id=1508219

Meet the Virts

Tom Killalea

http://queue.acm.org/detail.cfm?id=1348589

Mache Creeger ( [e-mail]) is a technology industry veteran based in silicon Valley. along with being a columnist for ACM Queue, he is the principal of emergent technology associates, marketing and business development consultants to technology companies worldwide.