Jim Frey is Vice President of Research, Network Management for Enterprise Management Associates (EMA) and is a featured guest blogger.
In order to remain engaged and relevant in the long run, a growing percentage of enterprise IT organizations are seriously considering transforming themselves into internal private cloud service providers. The requisite enabling technology is virtualization across all three primary technology domains – compute, storage, and network – combined with a healthy dose of automation and orchestration. Server virtualization is the most mature, and has taught us much with regard to automation and improving IT agility, fueling a desire to extend the model and achieve fluid architectures such as the Software Defined Data Center (SDDC).
While the technology behind SDDCs is rapidly approaching true viability, putting an SDDC into production requires a major shift in how IT is implemented, consumed and managed. Traditionally, the datacenter has been very device and hardware-centric, which required deep siloed expertise in each technology area. Moving to software-centric strategies introduces a layer of abstraction that paves the path towards truly service-oriented IT, up and down the entire stack. Instead of thinking of applications and services tied to a dedicated infrastructure, the model can shift towards IT as a provider/broker of hosting and connectivity services, upon which applications will reside and over which end users, customers, and partners will reach and utilize those applications.
The Growing Adoption (and Impact) of Cloud
EMA has been tracking adoption of cloud-based computing initiatives for the last several years. In a recent study, “Managing Networks in the Age of Cloud, SDN, and Big Data: Network Management Megatrends 2014,” EMA found that those reporting private cloud initiatives (43%) were slightly ahead of public cloud (39%) deployments. The study also confirmed that cloud initiatives are finally moving from early adopter status into the mainstream. Along with that growth, it is important to note that respondents reported cloud-related impacts on network management activities and practices that rose from 36% in 2012 to over 50% in 2014.
This impact growth is in part due to the fact that cloud adopters are having difficulty troubleshooting and monitoring external cloud deployments and are looking at the network as the last “jumping off” point. It also reflects the fact that software-defined architectures like SDDCs, perhaps the most promising approaches for implementing and capturing the true value of internal cloud transformations, create new barriers to visibility that will require deeper efforts on the side of network monitoring.
The Biggest Challenge – Network Virtualization Lagging
Successfully transforming IT infrastructure from a hardware-centric to a software-centric model depends heavily on the adoption of virtualization across the board, including compute, storage, and network. While server virtualization can be considered a mature, mainstream technology, the adoption of virtualization across other aspects of the infrastructure is more varied, with networking being the furthest behind.
Networking virtualization is in no small part hampered by the fact that multiple approaches are being championed from different quarters, both inside and outside the networking community. Overlays (e.g. VXLAN, OTV, NVGRE) are the darlings of the server admins, but cannot succeed without technology/protocol and capacity support of underlying network elements. Underlays (i.e. OpenFlow) are interesting for a host of other reasons, chief among them the ability to enforce consistent policy and to adapt readily to specific connectivity requirements, but are not ready for mainstream adoption. And controllers seem to be popping up all over the place.
At this point in time network virtualization is a piecemeal effort at best, with various initiatives and products at different points of maturity and none providing the complete answer to solving the network virtualization dilemma. This uncertainty has created a perceptible drag on SDDCs as a whole, and will have to be resolved if complete datacenter transformation is to be truly successful.