Virtualization and Cloud executives share their predictions for 2014. Read them in this VMblog.com series exclusive.
Contributed article by Dan Conde, Director of Products, Midokura
Wheel of reincarnation makes another turn in 2014
1. Shared network resources
means policies become important
We're back in an increasingly
shared-resource, time-sharing-like (aka cloud) world, so we will start to
recognize all of the implications in the networking and application
realms. Wheel of Reincarnation was a phrase used in computer graphics where
systems designs swing back and forth like a pendulum. A
similar turn of the wheel is happening due to cloud computing where we share
computer resources, much like in the time-sharing days, when people logged into
multi-user operating systems from terminals.
Instead of terminals, we now have use web browsers, RESTful API clients,
or whatever fits our needs, but the fundamental issue is the same: more
sharing.
This type of sharing was first
revived when virtualized x86 servers and with SAN/NAS storage arrays for
storage. The same type of wheel turning will affect networking, as it also
starts to become virtualized. Rather
than having dedicated VLANs or even physical network segments dedicated for
networks, a commonly shared physical underlay may be used in an overlay virtual
network. The implications are that
application owners need to be keenly aware of whether or not they are getting
the service levels from the shared pool of resources, and this is just like how
it was in the timesharing days, and this leads to the next point.
2. Applications are what
matters. Really.
Due to the increasing level of
sharing mentioned above, notions like Quality of Service, or Service Level
Agreements become extremely important in shared infrastructure
environments. This has always been a
concern in networking, but in 2014, people will realize that tying application
performance a policy to networking becomes very important. Initiatives such as Cisco's Application
Centric Infrastructure will bring these issues to top of mind for IT
managers. Since networking is the key
element that ties modern application workloads, people will look at the
infrastructure from the viewpoint of applications, and notions such as an Application Controlled Network, will
arise. A bottoms-up device & infrastructure centric view will not go away,
but will no longer be the center of gravity.
The proper view will be policies at the top, and mechanisms and
infrastructure at the bottom with orchestration in the middle, just as it was
in the old days. Some autonomy of
control will also become important since a centralized set of policies cannot
work in a large, shared multi-tenant cloud environment.
3. Software-Defined-Clarity
In 2013 we got overwhelmed with
Software-Defined-Something. That includes
the Software Defined Data Center, Software Defined Storage, and Software
Defined Networking. In the networking
arena, this became muddied as any equipment that has some form of
programmability and control was considered a form of software defined networking. Meanwhile, virtualized networks also got
lumped into software-defined networking.
I predict that in 2014, we will use a clearer definition - to separate
network virtualization (mostly overlay networks) vs. automation of physical
devices and people will stop mixing them up (although many will continue to do
so). This swing of the pendulum has
occurred in the past, and will affect the use of catch phrases such as SDN or
NFV, and people will start being more specific.
4. Network virtualization will move beyond POC and pilot stages
IT groups spent 2013 examining
network virtualization in a test-phase, but they will start to put serious
effort into putting it into production class environments. This is a simple prediction - and it's hard
to put numbers on the level of adoption, but the same cycle of curiosity,
skepticism, "show me", and "put into production" has occurred with adoption of
compute virtualization, and I predict will occur in network virtualization in
2014.
So there you
have it. It's a prediction of timing of
how the wheel turns. Anyone can be
right if you wait long enough, like a stopped (analog clock) being correct
twice a day, but the hard part knows how fast the wheel turns and when to throw
the dart into the wheel.
##
About the Author
Dan Conde is the Director of Products at Midokura,
a network virtualization company. Dan has held product roles at VMware,
Rendition Networks, NetIQ and Microsoft. Dan received his Masters of Business
Administration from the Haas School of Business, University of California,
Berkeley, and his Computer Science degree from the University of California at
Berkeley.