Lippis Report 195: CiscoLive, GigaOm Structure and IBM DOVE, Oh My

June could not have been a busier month for the networking industry. Cisco launched Cisco ONE, Cloud Connect and Unified Access at CiscoLive. IBM System Networking launched an SDN stack tied into the IBM PureSystems computing platforms of Expert Integrated Systems. And the folks at GigaOm delivered another great Structure conference that brought in cloud thought leaders. In this Lippis Report Research Note, we highlight some of the more significant June of 2012 announcements.

Siemens Makes OpenScape UC Mobile and Scalable in Version 7

Watch the Video

No IT firm organizes a user and analyst conference like Cisco, and this year’s CiscoLive did not disappoint. It was only one year ago that Cisco was being beaten up for having too many adjacent product lines and losing focus on its core networking business. Financial analysts and market watchers were calling for John Chambers to step down. What a difference a year makes. Cisco executed and, like a laser, focused on switching and routing, gaining share and innovating. At CiscoLive, Cisco launched three huge initiatives: Cisco ONE, Cisco Cloud Connect and Unified Access. We dive into Cisco ONE and Cloud Connect.

Cisco ONE: Cisco just launched an encompassing initiative to open up networking called Cisco ONE that offers programmability of the entire Cisco networking infrastructure, well beyond Layer 2 and 3 switches. ONE includes, and goes well beyond, the Stanford “controller-OpenFlow” Software-Defined Networking or SDN definition. OpenFlow support on select switches, a programming environment for network infrastructure, proof-of-concept controller for campus slicing, plus hybrid and virtual overlay network approaches are included in ONE. In short, Cisco is planning to offer multiple approaches to program networks across its portfolio stack from transport to management and orchestration.

World IPv6 Day Marks Massive Transition in IP Addressing: What It Means to You

Listen to the Podcast

Drivers for Cisco ONE are many, including the reduction of operational cost to manage networks by automated network configuration, the ability to evolve a network into a business enabler by becoming programmable and application-aware. There are three components of Cisco ONE: 1) OnePK, a network programming environment, 2) Campus Slicing or the ability to carve out a virtual network for researchers and 3) Virtual Network Overlaying.

OnePK: Only recently have networking vendors started to offer north bound APIs that expose certain aspects of their network operating systems and the rich information collected by network devices. Cisco’s OnePK offers programmability of the entire Cisco networking infrastructure; promising that OnePK will offer API’s for IOS, NX-OS and IOS-XR used primarily by service providers.

OpenFlow: The Next Generation in Networking Interoperability

Get the White Paper

Cloud computing providers, service providers and the largest of enterprises have been asking for these APIs so as to customize their networks, offer differentiated services and provide a framework that allows applications to request and configure network services. In short, OnePK creates a direct link between the network and applications, which is in stark contrast of today’s decoupled network-application relationship.

Campus Slicing: The traditional SDN model separates network control from data forwarding in an effort to reduce operational cost and add network isolation flexibility. A centralized controller programs an Ethernet switch’s forwarding table via an OpenFlow agent. This model was developed first within universities, such as Stanford, Berkley and Indiana U, as a means to partition or slice a network into multiple logical networks so as to provide separation between an operational network and a network used by researchers to perform experimentation.

Broadcom Smart-Buffer Technology in Data Center Switches for Cost-Effective Performance Scaling of Cloud Applications

Get the White Paper

In universities, slicing activities are primarily performed off the campus backbone, as the software used to implement both the operating systems and slicing functions does not provide policy management consistency required for production network applications. As a component of Cisco ONE, Cisco will be offering OpenFlow agents on select switches plus an SDN controller designed to operate in production networks that leverage policy, security and management.

Virtual Network Overlay: Virtual networks are logically isolated networking entities with unique attributes. In an SDN context, virtual networks may be overlaid on top of existing Layer 2/3 networks by creating tunnels such as VXLAN, NVGRE, potentially LISP, etc. These virtual network overlays are created in hypervisor vSwitches. Further as Layer 4-L7 network services, such as security, load balancing, WAN optimization, etc., migrate to the virtualization domain adding network services to virtual network overlays not only becomes easy, but the virtual networks gain attributes.

Cisco Cloud Connected Solution: Prepare Your Network for the Cloud

Get the White Paper

Vendors who own vSwitches are in an envious position as they have the means to innovate and control virtual network features and approaches. Cisco ONE utilizes the multi-hypervisor Nexus 1000V virtual switch to create virtual network overlays through programmatic interfaces to its virtual supervisor module (VSM), enabling automation-centric provisioning.

My Take: Cisco was quiet for the last 18 months as the industry discussed SDN and OpenFlow. With its Cisco ONE announcement, it has expanded the SDN industry discussion to programmable networking, thanks to Cisco ONE being the most extensive approach to programmable networking and SDN, thus far. Are there limitations and questions of openness vs. standardization? Yes, but Cisco has responded to a market shift in networking in an effort to catch up with compute innovation. In short, it is pivoting with the market.

Spring 2012 Open Industry Network Performance And Power Test Report

Get the White Paper

Cisco Cloud Connect: Cisco developed a framework and new ecosystem to enable a cloud connected branch office. There are four components to the framework, including Cloud-Ready Platforms, Cloud-Ready Network Services, Cloud Connectors and lastly, Integrated Management and Policy. Note that there are no new hardware platforms in the Cloud Connect Framework; all new components are software based to ease deployment at speed and scale. The underlying infrastructure is cloud ready.

Cloud-Ready Platforms: The Cloud-Ready Platform are Cisco’s Integrated Services Router or ISR G2 for branch offices, Aggregation Services Router or ASR 1000 for headquarters/data center plus internet access and the new CSR or Cloud Services Router for the cloud. The CSR platform is a virtualized version of the ASR IOS XE image. It will eliminate most traffic hairpinning/backhauling by delivering secure connectivity directly to the cloud from the remote sites. Hairpinning or backhauling is cloud or internet traffic that travels over the corporate WAN only to be re-routed at a data center to an internet connection.

Cloud-Ready Network Services: Network services are software that run on various platforms. Cisco has invested in a set of existing network services to assure they are cloud-aware in five key areas including Visibility for cloud applications, Optimization for cloud-specific applications, Security to eliminate hairpinning. As an increasing number of collaborative applications are now hosted in the cloud, a Collaboration network service provides increased access efficiency. Finally an Application Hosting network service brings the same operational efficiency and deployment flexibility enjoyed in the data center to the branch office or remote site by offering a more powerful hosting option in the ISR G2. The Application Hosting network service provides IT architects the design option to host applications close to users to increase user experience/performance and reduce MPLS/WAN north-south traffic flows.

Cloud Connectors: The Cloud Connectors are a new set of solutions designed to solve challenges for specific cloud services. Currently, there are four Cloud Connector categories, including Collaboration Survivability, Web Security, Storage and third party. These are but just the first set of Cloud Connectors from Cisco as there are many more in development at Cisco as well as partners since Cisco has developed a Cloud Connector ecosystem. This is an integral part of Cisco’s SDN strategy, where the rich intelligence of Cisco’s networking platforms can be harnessed through open APIs via OnePK for organizations to build their own software solutions to improve services to their customers, employees and partners. For example, third-party partner, Ctera, uses the OnePK API to offer a Storage Connector that provides secure access to cloud-based storage. Cloud Connectors are software modules that currently reside on ISR G2 platforms.

My Take: In its first Cisco Cloud Connect announcement, there are three connectors, a connector ecosystem, updated network services plus a new platform being launched. There is a new focus of investment at Cisco on inter-platform communication/collaboration engineered to deliver cloud-based services that improve user experience, increase cloud security and simplify operations. Therefore, all platforms will be participating in traffic statistic data collection to gain increased visibility and control of IT resources. All of the above can be summarized as the WAN is being re-invented to support cloud computing with Cisco Cloud Connect for the branch office.

IBM’s SDN Strategy and Vision for the Virtual Network of the Future

On June 28th, IBM detailed its SDN strategy with special emphasis on virtual overlay networking. IBM is like Cisco in that they both own vSwitches for VMware which enables them to control their future by innovating within and on top of this critical piece of technology that is key to SDN and virtual networking.  

IBM is investing in one of the most impressive SDN stacks that span physical and virtual switches to virtual network overlays that are configured via applications within the IBM PureSystems computing platforms of Expert Integrated Systems. For example, IBM System Networking has added OpenFlow to its 10/4OGbE IBM RackSwitch G8264. IBM’s vSwitch, the Distributed Virtual Switch 5000V (DVS 5000V), recently added to IBM System Networking’s portfolio, is an example of IBM’s investment toward a virtual network overlay switch. The Layer 2 DVS 5000V controller and Data Path Module for VMware create and manage virtual networks that are distributed over many ESX hypervisors. 

The DVS 5000V supports 802.1Qbg as does its physical switches, providing the basis for automated virtual network configuration and tunneling over physical network switches supporting both overlay and hybrid virtual networks. But as the number of VMs per multi-socket server increase tenfold every 10 years, IBM is investing in its Layer 3 Distributed Overlay Virtual Ethernet or DOVE networks to enable large scale virtual network overlays.  

In addition to physical and virtual switches, IBM System Networking was one of the first firms to demonstrate and work with NEC and Big Switch Networks to utilize their SDN controllers. Further as Layer 4-7 network services, such as security, load balancing, ADC, etc., migrate to the virtualization domain adding network services to virtual network overlays not only becomes easy, but the virtual networks gain these attributes. In the SDN stack IBM is investing in, applications build DOVE networks, which interconnect workload groups through a set of middle-boxes (i.e., L4-7 services). A DOVE Network provides the network middle-box connectivity associated with a “Virtual Systems pattern” that can be deployed multiple times with minimal manual configuration. 

IBM is unique in that its road map provides for the linkage of applications to call upon network resources, an advantage that very few firms offer. IBM’s new PureSystems family of integrated compute, storage and networking incorporate “patterns” today –the ability to configure an application once, such as an online ordering system, and then replicate as needs grow. IBM is investing in SDN to enable adding middle-box connectivity to these patterns. Terms such as clients, DOVE networks, Virtual System pattern, workload groups, etc., are used to describe how applications will call upon network resources in IBM’s SDN stack.

My Take: IBM is approaching networking from the top down, that is from a compute perspective in the data center. It’s focused on auto configuration of virtual networks and converged fabric too. It understands the key driver to SDN, and that is reduction of operational cost through auto-configuration; IBM’s SDN stack focuses on this end game. The above terms used to describe their SDN stack, such as DOVE, Virtual System Pattern, workload groups, middle-boxes, etc., will take some time for IBM to articulate to the industry so that they become widely known, and as a result, so too does its SDN Stack. 

2 Debates over Lippis Report 195: CiscoLive, GigaOm Structure and IBM DOVE, Oh My

  1. Why should Solutions & Applications Architects care about SDN? said:

    […] and see how it can be constructed with the help of a high-level API such as an Application Pattern Connectivity Service. This level of programmability gives Joe a powerful new tool to program and influence his network […]

  2. Why should Solutions and Applications Architects care about SDN? | the elastic cloud said:

    […] and see how it can be constructed with the help of a high-level API such as an Application Pattern Connectivity Service. This level of programmability gives Joe a powerful new tool to program and influence his network […]