GigaSpaces is a pioneer in software middleware for large-scale real-time applications. Its Cloudify open source framework brings web scale to cloud orchestration and automation. This TOSCA-based solution addresses what’s happening today in hybrid cloud and network functions virtualization.
NFVzone recently interviewed GigaSpaces CTO Nati Shalom to talk more about the company; NFV adoption, benefits, and challenges; and what he expects in this arena in the year ahead.
Tell us more about what Cloudify.
Cloudify serves as the NFV Orchestrator and Generic Virtual Network Functions Manager in the context of the ETSI MANO architecture. Cloudify orchestrates and manages virtualized and non-virtualized infrastructure, devices, and applications, as well as OSS and BSS components, in a brownfield environment, through a single pane of glass.
Cloudify enables VNF providers, carriers, and vendors alike to realize the benefits of the cloud by assisting in the transition of non-virtualized physical appliances through virtualization, and finally transforming them into cloud-native virtual network functions with full lifecycle management and orchestration from the deployment phase through intelligent placement awareness – and leveraging the capabilities of the underlying infrastructure (DPDK, NUMA, SR-IOV, etc.), and service function chaining during runtime, including day 2 operations such as self-healing, auto-scaling, and teardown.
Where are we with NFV adoption?
With the ETSI NFV architecture spec, we saw the first wave of opportunity and real adoption in this space. Carriers and vendors alike still use that architecture as the basis for building their NFV product lines. However, we are really still in the baby stages of actual, production-grade NFV. With the open sourcing of AT&T’s ECOMP architecture project, we are seeing the next phase of adoption and carriers potentially taking the holistic approach to NFV.
How – and where in the network – are service providers proving in NFV?
With NFV, when you chain together physical network functions and virtual network functions, and make the VNF chaining dynamic, VNFs can be added or removed on-the-fly. For example, if you have a basic connection from the CPE to the cloud, you can provision a virtual firewall in the cloud to add security. You can also provision URL filtering to block URL content or deep packet inspection for traffic analysis.
Being agile and dynamic with modular components like Cloudify, you can build and change services based on SLAs or end user requirements. This wasn’t possible before virtualization and, more importantly, orchestration of the network functions.
To name some specific scenarios, vCPE and SD-WAN are the most popular solutions being offered, and driven, by service providers. Additionally, mobile core components such as vEPC, SBC, and vIMS are also very common.
What are some of the biggest challenges to NFV implementations?
Culture change is now the biggest hurdle to faster NFV adoption, consolidation between the carrier network operation and IT operation. As carriers adopt cloud practices to run their core infrastructure, the need to have separate groups and frameworks to manage their IT and network services becomes obsolete. This will lead to a major culture change in those organizations and will become the biggest barrier to NFV adoption.
Another challenge is lock-in, as operators have a tendency to work with turnkey solutions rather than build a stack that allows for portability of applications as well as interoperability between clouds.
There are a lot of industry groups addressing NFV MANO. Who's leading the charge?
There are currently three main projects.
ECOMP - Led by AT&T, this project, according to their site, ‘is the brain providing the lifecycle management and control of our software-centric network resources, infrastructure and services.’
Open-O - Open-O is a Linux Foundation open-orchestration project with participation from China Mobile, ZTE, Huawei, China Telecom, Ericsson, VMWare, Intel, PCCW, and GigaSpaces/Cloudify is proud to be a founding member of this project as well.
ARIA - ARIA is an open source, open governance library of orchestration tools for TOSCA orchestration led by Cloudify. (It’s currently used by Open-O, soon OSM, and others that can’t be named.)
What is TOSCA?
TOSCA stands for Topology Orchestration Specification for Cloud Applications. It is a standard modeling language, defined by OASIS, which is used to describe the topology of cloud applications and services to ensure interoperability between clouds. It is currently being used by many of the NFV MANO Orchestration services.
How does what's happening in NFV relate to virtual machines?
Many of the virtual network functions run as Virtual Appliances. The performance characteristics of those network functions drive specific performance optimization such as SR-IOV, DPDK, etc.
How does what's happening in NFV relate to bare metal?
Bare metal cloud provides the agility of cloud without the hypervisor overhead. Hence it is commonly used to run high performance VNFs.
How does what's happening in NFV relate to white boxes?
White boxes are used as low cost devices to run VNFs. They are commonly used to drive vCPE, specifically low-cost edge devices.
How does what's happening in NFV relate to containers?
Containers are a popular choice for packaging and delivering cloud native VNFs. Containers also come with a rich ecosystem of tools such as Docker Swarm, Kubernetes, and Mesos. Containers are enabling a new generation of microservices-centric and high-density NFV architectures that will revolutionize the way that network services are delivered.
How does what's happening in NFV relate to DevOps?
Continuous deployment is one of the most common DevOps practices. Cloud native VNFs and NFV use DevOps methodologies and practices for handling continuous upgrades of NFV services as an alternative to the previous services-led and manual upgrade processes.
How does what's happening in NFV relate to cloud service providers like AWS and Microsoft Azure?
Cloud service providers are starting to offer a new class of performance-centric services such as bare metal clouds and a marketplace of network services that make the deployment of NFV services on public cloud possible. Currently, public clouds are used mostly as access capacity resources to extend the private cloud resources.
What are your expectations for NFV in 2017?
1. In 2017 we will see greater adoption of vertical NFV deployments such as SD-WAN, vCPE, and vIMS.
2. Many of the traditional VNF providers are going to be disrupted by new startups that are coming with cloud native VNFs.
3. We will see consolidation of the open orchestration initiatives such as ECOMP and Open-O, which will further disrupt the carrier market.
4. Traditional network vendors such as Cisco and Juniper will have to go private (similar to Dell) due to the business model disruption that makes their current model unsustainable in a post-NFV world.
5. NFV and DevOps will converge based on the work done by the telecoms, and make a move into the enterprise. This enterprise NFV shift will benefit enterprise operators of large, global networks and hybrid cloud deployments. Taking lessons from NFV, enterprise hybrid cloud and DevOps will be seen as the path forward.
Executive Editor, TMC
LG U+ is using the NFV-based route reflector feature of the Juniper Networks vMX and EX4300/EX2200 Ethernet switches. This new installation puts the K…
Amdocs has signed on to be part of the Linux Foundation's ECOMP effort. ECOMP stands for enhanced control, orchestration, management, and policy.
CenturyLink gets a little more life out of a Department of Defense contract by bringing in both SDN and NFV capabilities.
Netcracker Technology made a strong showing in new NFV MANO research published by Current Analysis. The good news is that Netcracker was named the ove…
Saguna Networks turns to Wind River's Titanium Server for NFV infrastructure.