Thank you to everyone who joined SDxCentral for its latest DemoFriday featuring Accedian Networks on performance monitoring. Attendees were able to learn how they could fully benefit from software-defined networking (SDN) capabilities, and at the same time, preserve and build on their existing networks. During an interactive demo session, the audience saw the Accedian virtualization lab and shown how companies were moving towards total quality-of-service (QoS) and QoE. Following the presentation on performance monitoring, Scott Sumner and Greg Spear kindly responded to questions from the audience.
Accedian: No experience is necessary. Despite the use of NFV principles to reduce the vCPE hardware to the absolute minimum, the VCX Controller virtual appliance uses the same user interface, and supports the same CLI command set as its physical appliance (NID) predecessors. The VCX can also be delivered on a physical appliance. However, the VCX can also be run as a virtual machine on KVM or XXX, and can be orchestrated by a variety of open source and third party orchestrators.
Does the solution support distributed NFV (D-NFV)?
Accedian: Yes, multiple VCX Controllers can be installed throughout a network to support localization requirements where NFVI hosting certain performance assurance functions requires local processing or very low latency response – e.g. packet capture to feed to a locally installed DPI appliance. Generally, though, the VCX is centrally installed in a data center. As time stamping for all performance assurance functions is performed at the monitoring point by the nano smart SFP or the ant module, telemetry delay is does not affect delay measurement precision.
How would existing management systems interface with the virtualized solution, compared to the physical NIDs?
Accedian: External platforms can communicate in the same way with the VCX Controller as they would with a standalone NID – SNMP, CLI, open-API or by using a web GUI. Each module is seen as a remote port, so that element management systems require only one connection to a VCX to manage all modules under its control. This approach is highly cost effective from an NMS/EMS licensing perspective. The familiar communication options makes migration from physical to a hybrid or fully virtualized instrumentation/vCPE environment straightforward, as operational procedures can be maintained without modification, in most cases.
What are the cost benefits vs. alternatives?
Accedian: vCPE and monitoring solutions using the VCX and Accedian Performance Modules in place of standalone NIDs or routers can be up to 90 percent less expensive, depending on configuration and scale. Accedian has an ROI tool to help service providers calculate their savings.
Can this solution use existing network elements as test endpoints, instead of requiring the smart SFP?
Accedian: Yes, as the VCX Controller and modules support standards-based performance assurance protocols like RFC-2544 and Y.1564 for service activation testing (SAT), and TWAMP and SOAM (RFC-5357 and Y.1731) for performance monitoring, they can interact with existing infrastructure supporting these standards, using these network elements as test reflectors instead of requiring modules or NIDs to be installed.
How many service providers have actually deployed this? Is it available now?
Accedian: To date, seven of the largest service providers globally have installed the solution in scale. Over 100 service providers in more than 25 countries have deployed in trials or regional installations.
Do these physical and virtual solutions work together?
Accedian: They work seamlessly together, using the same EMS and management methods to ensure operators can migrate to a virtualized CPE method while maintaining their existing installations, as well as choosing a hybrid environment where benefits of this approach are important (e.g. using NIDs for multi-tenant connections, and smart SFP modules for direct fiber connections to an enterprise).
How far can the controller be located from the smart SFPs?
Accedian: As there is a patented ‘lossless’ telemetry tunnel setup between the VCX Controller, and the modules under their control, distance is not an issue. A reliable connection with “reasonably” low packet loss is required – any standard management VLAN connection would suffice. Time stamping is locally performed by the module, and communication overhead is minimal, so centralizing one VCX per market or region is feasible, depending on the number of sessions running (e.g. TWAMP monitoring sessions, SAT sessions, etc.).
Accedian: This solution can and is used for both. The standards employed and the resilience designed in follow telecom service provider guidelines, and are certified and hardened for large scale telco networks. The control signaling is also optimized for large geographic areas, making this solution well suited to large-scale WAN deployments, which can be more demanding from an instrumentation and vCPE point of view than a local deployment within a data center. The precision of the monitoring, however, can be used to monitor very short links and substantially identify minute sources of delay and congestion. In short, the solution offers the precision and small footprint required for data center applications (LAN), with the telecom-grade design required for WAN applications.
Is the nanoNID getting the same/similar capabilities as the VeloCITY series?
Accedian: Yes, the ultra-fast (100 µs) microburst detection, time stamping, packet brokering capabilities will be made available in the second generation modules which are now available.
Does the VCX have an API to integrate with non-vendor orchestrators?
Accedian: A number of API options are available, in addition to a REST-ful interface, including NetCONF/Yang for SOAM configuration (and expanding to service definition). Please contact Accedian for complete details.
Can you share any more details about the lossless tunnels between the VCX and modules?
Accedian: The tunnels are TCP-based, but employ proprietary, patented methods to compress bandwidth requirements, while also ensuring the lowest-latency transmission of data between modules and the VCX Controller. The suitability of a network connection is verified when the tunnel is established, to ensure that any connectivity issues are identified before a session is permitted to initiate.
Is the customer traffic encrypted?
Accedian: The solution does not alter or inspect the payload of the customer traffic in any way. This leaves the service provider many options to perform encryption using their preferred technique (MACSEC, IPSec, etc.) with virtualized or localized encryption methods. Please see our white paper, “vCPE – Performance Assured” for more details.