The first NFV MANO (Management & Orchestration) solutions are now moving from the proof-of-concept (PoC) phase into production, and the focus of attention is shifting to mundane but critical issues of service assurance, reliability, availability, and manageability of the solution.
Customers expect NFV solutions to provide the same level of dependability they’ve been accustomed to with their legacy solutions. But what if something goes wrong? What if there is a failure? Who is to blame?
Nobody is perfect, but not being prepared for failures is inexcusable. That is why we’ve released 3 mini-white papers that address how these questions can be answered in an NFV MANO environment.
Root cause analysis
Root cause analysis is nothing new in the networking industry. But NFV presents new challenges due to its multi-vendor layered architecture, which separates applications from the infrastructure on which they run. Is a problem due to a fault in the application software, or is it from a failed server, network interface card, or misconfigured switch? How can we quickly pinpoint the source of a failure?
In order to determine a root cause, 1st of all we need to be alerted to failures, to the right failures. This is not self-evident in NFV solutions composed of multiple application-specific and generic functions coming together to deliver communications services.
For example, failures of underlying resources such as a switch failure or high server load may not be reported to NFV applications and they may not even have an immediate impact. From an application point of view everything might appear fine, but there may actually be significant risks to the applications. How can we make sure that risks are identified and protective measures are taken when it is still possible?
Being able to identify and report failures is not enough, because repairing the failure after it has been reported may just take too much time.
Service providers know they need to maintain their high standards of service availability. But the magical -48V remote power supply is no longer available to achieve high availability. And NFV solutions pose extra challenges since we are dealing with off-the-shelf hardware elements that – taken individually - may not be designed for the same carrier-grade standards as traditional networking “boxes”. In addition, we are introducing extra layers such as infrastructure and additional management functions, any of which can fail.
To solve these issues, NFV relies on availability concepts from the cloud but the OpenStack® open source software, serving as the foundation for virtually all NFV MANO installations, is not always ready for what is needed. How can we provide high availability in the short and medium term and how will HA solutions evolve as NFV gets deployed broadly?
I still remember the excitement at the SDN & OpenFlow World Congress 2012 when the initial NFV whitepaper was published. When 13 service provider authors threw down the gauntlet and challenged the industry to adopt and adapt virtualization and cloud technologies from IT to the world of networking.
Many at the time doubted whether that was ever going to be possible given the enormous performance and reliability demands commonplace in networking. And given the almost 100% dependence on custom hardware.
On the other hand the prospect of unprecedented agility and significant cost savings despite a market of rapid traffic increase, price per bit decline, and non-traditional competitors caught the interest of everybody.
With NFV solutions, we could:
- Deploy network functions anywhere at any time, without sending staff on location
- Eliminate long term capacity planning
- Reduce cost by using off-the-shelf hardware
- Take advantage of new pay-per-use pricing models
NFV solutions today
By now PoCs have shown a lot of the NFV MANO promises to be feasible from a technical perspective. ETSI has conducted some 38 PoC demonstrations. We have shown automated, policy-controlled placement of workloads. We have shown high throughput packet I/O, automated deployment, scaling, and healing. A lot of fancy stuff.
And now it’s time to meet customer expectations in a production environment. Download our 3 white papers for a closer look at what’s in store.
NFV solution web pageCloudBand solution web pageMotive Dynamic Operations solution web pageReshaping the future with NFV and SDN: A Report by Bell Labs and Arthur D. Little
To contact the author or request additional information, please send an email to firstname.lastname@example.org