White Papers

Understanding OPNFV

Issue link: https://read.uberflip.com/i/1344850

Contents of this Issue

Navigation

Page 44 of 144

Understanding OPNFV 44 Procurement Models There are many sub-variations, but broadly speaking the models are: ● 100% do-it-yourself (DIY): In this model, the user purchases commodity hardware and utilizes only open source software. This results in staffing and, more importantly, retention of in-house software engineering, integration and operations. A substantial investment is required to create architecture, engineering, testing, DevOps, integration, interop testing, bug fixing teams and ops support as well. Furthermore, operational teams are required for design, deployment, monitoring, lifecycle management, break/fix support, and so on. In addition to investment and a level of sophistication, the user would also need the scale to amortize these investments. Ultimately the question to ask is, does a 100% DIY approach provide any differentiation, or is it a distraction? ● Software from different vendors: In this model, the user purchases commodity hardware and procures software from a variety of vendors (proprietary or open source). The user would act as a system integrator and then own operations as well. The burden is less than a 100% DIY case, but still substantial. DevOps, integration, testing, interop testing, bug fixing would still be the user's responsibility along with all operational tasks. The user would have to be adept at finding issues in a multi-vendor environment where finger pointing amongst vendors could kill months. ● External system integrator: In this model, the user would purchase the entire solution from one vendor, yet retain operational management of their cloud. While less burdensome that the first two models, operating an NFV cloud would be the user's responsibility. ● Managed services: In this last model, the user would purchase the solution from one vendor and then ask them (or another vendor) to manage it. This is the least burdensome option from an effort point of view, but may be more expensive at scale. Clearly, there is no right answer in terms of these options. The answer may be different for different users. Choosing the wrong model could actually add to the TCO rather than cutting it.

Articles in this issue

view archives of White Papers - Understanding OPNFV