- When public cloud computing services first emerged several years ago, little focus was placed on the network, with the initial value proposition leveraging the Internet for connectivity between virtual machines hosted by providers and their end users. Flexibility, availability, and affordability drove early market adoption by user groups which place a high value on those service attributes.
- Fast-forward to 2013, and widespread adoption of cloud is coming closer to reality. Widely felt concerns about security and stability are being addressed in infrastructure and software as a service (IaaS and SaaS) solutions being rolled out by network service providers especially. Increasingly, service providers are positioning the network as a core component of their cloud solutions, citing advantages from managed connectivity to integrated network functions.
Cloud based applications are only as strong as the connectivity they run over. The network-centric cloud is increasingly being positioned against the ‘best effort’ server-centric cloud model because it delivers network-based security, strong SLAs and enterprise-class performance.
Globally, providers like AT&T have emphasized the “network enabled cloud” for some time, while Verizon is the latest to use its managed network to address performance, management, and security issues in public and hybrid cloud services. In Europe, Interoute’s VDC IaaS service, for example, leverages the fabric of the network to create virtual data centers that integrate computing power with network resources. Its API automates the MPLS core, allowing cloud computing functions to leverage network resources in real time. Uniquely, the service can create any relationship between the VLAN of the computing and the WAN equivalent on the MPLS network (i.e., the VRF). For the customer, this means any physical data center architecture, corporate IT environment or major SaaS platform can be replicated automatically, online and in real time, and with exactly the same level of performance and security that is in place with a dedicated capability. Interoute is working on additional integration of network and compute which will allow any IP address to become part of the enterprise WAN, enjoying VDC-hosted IT services globally.
Orange Business Services has also been positioning the network as a differentiator since before it launched its Flexible Computing IaaS portfolio. Its current messages, e.g., “the network is the platform”, and “no network, no cloud”, are backed up by network investments of over EUR 700 million per year. Orange believes that CIOs won’t transfer ownership of apps and hardware to third parties without the kind of strong SLAs and QoS that only a managed network can provide. To that end, its Business VPN Galerie was positioned as a cloud connectivity service and has won industry accolades as the first service launch that leverages the network as a crucial element in cloud service delivery. The Flexible Computing solutions provide joint service management of network, infrastructure, and OS/applications, each with its own SLA but wrapped together in an overall SLA.
BT Global Services is bundling network and cloud (compute, storage, apps) in data centers across the globe, differentiating its offerings from non-network owning competitors by providing seamless integration with its global private networks and network-based security services. It sees this as essential in marketing its vertical specific cloud enabled solutions (e.g., BT Radianz for financial trading, BT 4 Life Science, and BT Global Trace for logistics) due to inherent requirements for security and latency.
Interoute is talking about collapsing Layer 2 and Layer 3, providing a broader set of options for integrated compute and network management and configuration, developing a third generation version of its VDC which leverages an integrated VPN, compute and communications node at the edge of the network, enabling NVF and function convergence in a new, complementary infrastructure layer.
Non-telco cloud providers AWS, Rackspace, and Google obviously have less to say about such matters, positioning instead on massive capacity, widely used APIs, and increasingly lengthy menus of instance and resource based service options. But it’s not a wholly black and white dichotomy between the two camps. Even a global network provider like BT, which trades on its ability to control service delivery and provide meaningful SLAs, sometimes must rely on third-party networks for customer access and connectivity in certain regions. Meanwhile, the non-telco Google owns significant network assets, enabling every compute instance, regardless of region and zone, to communicate to every other instance over a project private network.
How do you see the importance of the network in cloud computing? Do these telco innovations make the cloud more attractive for delivery of mainstream IT resources?