Provided by CRU
Thirty years ago, if you announced to your colleagues that you planned to put critical assets in “the cloud,” they would have questioned your judgment, or your sanity. Now, such a statement is fully acceptable and fairly well understood. Although the name and diagrams may suggest otherwise, the cloud is based on massive buildings, acres of floor space filled with racks of electronics, huge cooling systems, backup power supplies, and hundreds of kilometres of cabling.
The concept of distributed computing goes back at least 50 years. It pre-dates the PC. The term “cloud computing” is attributed to a 1997 lecture given by Dr. Ramnath Chellappa, a professor of economics and information systems. The term has been in general parlance for about 10 years. But the concept of a “cloud” has been used in telecom diagrams for many years to illustrate boundaries between dedicated and shared infrastructure, or to illustrate the Internet.
Many technical groups and publishers offer definitions of “cloud computing.” They vary; some are long, short, broad, inclusive, or very specific. Plus, there are different types of cloud services. The US National Institute of Standards and Technology, a government-funded lab, published a three-page definition in September, 2011. This definition has been widely cited. It begins as follows:
”Cloud computing is a model for enabling ubiquitous, convenient, on-demand network access to a shared pool of configurable computing resources (e.g., networks, servers, storage, applications, and services) that can be rapidly provisioned and released with minimal management effort or service provider interaction. This cloud model is composed of five essential characteristics, three service models, and four deployment models.”
Key elements of the NIST definition are that a cloud resource is shared among multiple users, it is easily accessed using standard network interfaces, and the services can be readily provided or expanded with new customer requests or with increasing demand. In business terms, the cloud is a way of outsourcing data storage, data processing, and other computer or network functions. The facilities that provide cloud services are at remote locations, and most people use the term to refer to services that are accessed or provided via the Internet. The billion-plus cloud users include both businesses and consumers.
The remote physical facilities can be across an ocean, across the street, or any distance between these two extremes. In any event, the users’ data is carried on telecom networks – usually on the Internet. The increased use of cloud services is a major contributor to the rise in Internet traffic, which is why telecom carriers must continue investing in network capacity.
Cisco Systems has been publishing semi-annual statistics on Internet traffic for six years as part of a service Cisco calls its “Visual Networking Index.” According to this data, Internet traffic has increased with a compound annual growth rate (CAGR) of 46% from 2005 through 2012, and it has more than doubled every two years since 2006. Cisco expects Internet traffic to grow at more moderate rates in the next five years, with a CAGR of 23% from 2012 to 2017.
Beginning in 2011, Cisco began tracking traffic associated specifically with data centres, including cloud data centres and what it calls “traditional” data centres. Cisco says cloud data centres provide public cloud services, which are available to multiple end-users with “elastic and scalable provisioning and usage-based pricing,” and which are delivered on demand. Cisco’s data shows that the total traffic for cloud and traditional date centres is growing at rates comparable to Internet traffic. In 2012, Cisco projected data centre traffic to increase with a CAGR of 31% from 2011 to 2016. In 2012, cloud data centres accounted for 39% of all data centre traffic, but this percentage will increase to 64% in 2016, with faster growth expected for cloud traffic than that of traditional data centres. According to Cisco, cloud data centre traffic will increase with a CAGR of 44% from 2011 to 2016, compared with a CAGR of 17% for traditional data centres.
In addition to the two-part segmentation of cloud and traditional data centres, Cisco’s statistics also show data centre traffic in a three-part segmentation as follows:
The first segment corresponds to the Internet traffic assessed in Cisco’s other information service, the Visual Networking Index. The VNI measures and forecasts Internet traffic based on end-user applications. This is mainly traffic between end-users and a data centre of either main type. There is a small amount of Internet traffic that does not originate or terminate in a data centre, such as data in some peer-to-peer applications. According to Cisco, the data-center-to-end-user traffic is only 17% of all data centre traffic. Another 7% is between data centres, and the vast majority, 76% is within a data centre.
The external traffic, from data centres to end-users, includes the data transmissions for web sites, email, social networking, many smart-phone “apps,” and video services. Combining Cisco’s data on videos downloaded by consumers and “managed IP” video, which refers to IP TV services, video files accounted for 54% of Internet traffic in 2012, and this percentage will increase to 63% in 2017. Video traffic is widely considered a major factor in Internet traffic growth and bandwidth requirements.
Yet Cisco’s forecast is that traffic within data centres will grow at a comparable pace. This internal traffic results from data centre architectures, which include links between application servers, storage systems, and databases. The internal functions also include redundancies for reliability and backup storage. Cisco acknowledges that external data flows are expected to rise with increasing use of video services. But Cisco says that increasing use of virtual desktops and cloud services will rely on multiple data centre racks and services, and therefore cause internal traffic to grow as fast as external traffic.