In many sectors, IT departments work with a flat investment budget and declining workforce. At the same time, they are facing a growing demand for applications, storage and data analysis.
Faced with this reality, the Software Defined Network is interesting but only to the extent that it offers an immediate business value.
At this stage of its development, so-called northbound applications for SDN are more a promise than a reality. But the downstream applications , linked to network management, become deployable.
In the retail sector, data stores generally increase by 20 to 40% per year, and of course nothing is ever deleted. Big Data analysis with Hadoop can manage but also analyze this data. But, in the absence of effective orchestration tools, for both servers and networks, the development or installation of new Hadoop clusters can take anywhere from days to weeks.
Network, server, and application management teams must collaborate to deploy and test new types of installations.
On the server side, systems such as OpenStack can better manage such orchestration. On the network side, the SDN and its associated tools can also do this work. Reducing the time and improving the resulting accuracy of deployment are great advancements for the IS.
Other top-down applications that offer significant potential for enhancements include SAN storage , telephony and data convergence, and the unified network .
On the storage side, we are seeing a strong push for IP storage, both FCoE and iSCSI . Hence the need to allocate resources at a time depending on the capacities of the storage arrays concerned and applications in use.
The dynamic nature of virtualized servers requires that the network be equally responsive to the needs of applications. For convergence, it is of utmost importance that parameters such as quality of service and bandwidth allocation are easily configured and reconfigured to meet the needs of users.
And this is naturally the case in software management systems such as SDN.
Finally, one of the major weaknesses of the current network management systems that remains to be solved is the ability to aggregate usage data to present the overall network usage per application.