If your are a Head of IT department or you are managing your CSP's private cloud computing, a lot of questions, requests are coming for you soon.
Many Telco vendors are deploying their applications on top of X86 platforms and they announced even supporting virtualized environment.
In most of Telecom operators, IT and Telco network are separated activities, dedicated budget and teams.
Telecom vendors use to offer the entire solution from hardware to software to service.
Most of work was about integration new service within existing platforms, configuration and do planning and dimensioning of your network.
With the huge decline of Telco revenue and the success of OTT business model, many CSPs are thinking about reducing CAPEX and OPEX and trying to be me more efficient, innovative and more faster to reach a better time to market.
On the other hand, Cloud Computing is demonstrating a good success on IT domain.
One decade was enough to bring IT on top of resource revenue for big companies.
His elasticity, agility and Pay-as-You-Go model had demonstrated a new efficient way of doing business.
Many OTT players have adopted a Cloud based solution and have made a very good figure.
Even virtualized, these Telco applications have a specific requirements to consider when designing and preparing virtualized environment for them.
For IT managers and engineers, I invite you to take a deep look on ETSI initiative for NFV.
You will heard a lot about ETSI NFV and the right way is to understand the standard before starting any design for Telco applications.
NFV, VIM, VNFM, MANO, NFVI, EMS are words that will made your day.
ETSI web site (link) explains all of these standards.
There are specific requirements to get the virtualized network to carrier grade capability.
- Availability: Telco domain require at least five 9s of even six which means only few seconds of planned or unplanned downtime per year. Most of IT entreprise are about three 9s of availability.
- Performance: a large number of transactions happening in parallel.A large amount of throughput and lower latency are required by Telco applications. Hypervisor should handle a huge number of real time transaction without impacting quality of service.
- Security: Consumers should feel secure when they are transmitting their Data running on top of shared infrastructure.
- Serviceability: The network life cycle should be more longer that on IT domain. Network should be serviceable over a long of time, maybe more than 8 years.
- Service Continuity: Carrier want to built geo redundant network to assure disaster recovery and business continuity of network service in case of disaster or incident.
Real time applications need determinism almost the time.
Multi-tenancy and isolation are mandatory on your your virtualized Telco enviornment.
The NFV component VNFM is responsible of applications lifecyle management in term of intstantiation and auto-sacling feature. This component is dealing with your Virtual infrastructure Management VIM to assure that. Isolation and Multi-tenancy are the unique guaranty of boundaries between all of this applications coming from different vendors.
This amelioration done at the hypervisor layer to support carrier-grade applications made a lot of trade-offs:
- It's hard to jointly optimize for latency, throughput, availibility and efficent.
- Difficult to maintain virtualization benefits while optimize performance.
Some customization may also impact virtualization operational mode.
A lot of challenges are coming with the virtualization of the Telco network functions.
Service chanining, orchestration and Software defined network are other features that will help us reaching a Telco Cloud domain.