Where will Software Defined Networking be in four years?


Predicting things out too far is always risky – things you expect to happen quickly, take much longer to gain wide adoption; and things you expect to happen slowly, get boosted by “killer apps” that get adopted much sooner.  Sooo…with that in mind…here’s my view.

From my vantage point, success doesn’t mean that today’s networking model goes away.  In the Data Center, many disruptive and discontinuous technologies have lived alongside existing technologies, though they have changed things.  For example, high-end servers adopted CMOS and Linux, while maintaining all the attributes demanded by Enterprise clients, such as Continuous Availability, Security, Ease-of-Use and Performance.

So will the traditional network switch model, with a distributed control plane across the switches, go away?  No, but it will run alongside SDN’s 3+1 planes.  That is, our goal is for Open Daylight to provide a common, open source code base that can control the link, data, control and overlay planes, through open APIs.

In my view, Open Daylight accelerates the adoption of SDN, much in the same way as Linux accelerated the adoption of an open source alternative to proprietary OS in the x86 market.  The rate of acceleration will be determined by the client value of Open Daylight Applications – the higher the value, the faster the adoption.

So which applications have the most value?  To the clients I work with, a Connectivity Service that automates the provisioning of multi-tier systems, including all the linkages between compute tiers & network appliances, is in high demand.  A Connectivity Service automates what today takes operational teams days to weeks to make those linkages.  That type of response time is not acceptable for spot marketing use cases, where entire multi-tier systems need to be brought on-line rapidly to meet demand spikes.

As an example of the Connectivity Service’s value, consider a spot marketing use case where a Superbowl ad garners significant interest, resulting in social media spikes and analytic engines detecting the spike.  These analytic servers trigger a workload orchestration engine to provision many web VMs and associated security appliances, which provide additional post-game video commercials and follow-up target marketing, such as discount coupons through social media.  The workload orchestration engine also provisions multi-tier transaction systems to prepare for demand spikes, which as they occur, would also pass demand to the business’ supply chain.  All of this needs to happen in minutes to hours, not days.

Connectivity Service Diagram

Another example is increasing the intelligence of the workload placement engine.  Today, servers and storage provide an API ecosystem for real-time feedback, such as CPU utilization, IO utilization and disk access patterns to the placement engine, which is used to determine where a workload is placed.  Unfortunately, without network feedback a new workload may be placed on servers that connect through a switch that is already running elephant flows, which when combined with the new workload congest the network.  Providing an API ecosystem to the network allows the placement engine to take existing and projected network load into account before a workload is provisioned on servers.

From my vantage point, the main value of SDN will be the ability for orchestration and management applications to leverage network programmability, which as stated above has existed for years on servers and storage.  Just like servers and storage used this capability to better meet client and application Service Level Agreements, the main value for SDN will be in doing the same for the network, which will require both the overlay and underlay to be controlled through APIs by the Open Daylight controller.

In summary, Open Daylight provides a controller base that enables the types of applications described above, which provide a reduction in network operational expenses and, more importantly, faster response time to changes in business needs. The savings in operational expenses can be used to go after additional business opportunities.

Smarter Computing Analyst Paper - HurwitzTo effectively compete in today’s changing world, it is essential that companies leverage innovative technology to differentiate from competitors. Learn how you can do that and more in the Smarter Computing Analyst Paper from Hurwitz and Associates.

Subscribe to the Smarter Computing Blog

Recent Posts

Creating a fast-track for the hybrid cloud

Setareh Mehrabanzad

Last month, IBM Systems unveiled new solutions for creating an agile hybrid cloud architecture by enabling VMware’s vRealize Automation Platform for IBM Power Systems and IBM z Systems. Today, IBM Systems and VMware are introducing expanded capabilities this week at VMworld 2015 Europe in Barcelona.

Continue reading

Introducing the all new Power Systems LC Line of servers

Doug Balog

IBM has furthered its commitment to powerful and cognitive systems of insight, unveiling a whole new Linux class of IBM Power Systems designed for clusters and clouds: the LC Line of servers. The all new LC Line of Power Systems represents a different way to experience Power Systems.

Continue reading

Leave a Reply

Your email address will not be published. Required fields are marked *

* Copy This Password *

* Type Or Paste Password Here *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>