Content is driving network traffic growth at breakneck pace – and data centres are at the epicentre of this trend. According to a recent study, network traffic will grow to 15 zettabytes by 2020 with a CAGR of over 27%. As new technologies and applications come online, traffic volumes will only intensify and could even surpass these estimates over the next 2 to 3 years. Along with this data growth, data centre managers also face challenges such as interoperability, complex multifibre infrastructure and rising costs. Changes are afoot that require a new mind-set in the realm of test and measurement to ensure 24/7/365 uptime.
For operators, the network is their product and they conduct extensive tests – and more tests – almost every time they introduce a new service. Internet Content Providers (ICPs) have a different mind-set. Their product is content and the network is the vehicle to reach the user. Often ICPs conduct limited tests prior to the launch of a new service – they find it time consuming and a constraint on the pace of their growth. Understanding the ICP mind-set and its impact on network infrastructure along with their requirements is fundamental to the continued rapid expansion of data centres.
Increasing automation
Usually, telecom operators have hundreds of engineers to manage the network and most are hardware proficient. They consider the lifespan for hardware and technology to be around 10+ years. ICPs, on the other hand, have much smaller operations teams and their forte is typically routing and software. As such, they create open Application Programming Interfaces (APIs) and software-based automation to maximize workloads and networks.
What’s more, the phenomenal growth rate of ICP infrastructure requires them to rip out and replace technology every 3 to 5 years, so they view hardware with a limited lifespan. ICPs often find that the standards bodies move too slowly to meet the needs of their business model. As such, they frequently white box technology from multiple vendors – often before industry standards have even been agreed.
Fact is, ICPs have experienced rapid growth and need to be agile to operate in an ultra-fast paced world – yet for data centre infrastructure, all this can create problems of interoperability and downtime. If those challenges were not enough, data centres also have to buckle up for faster network speeds.
Maximize speed but reduce power
Speeds at data centre interconnects (DCIs) and intra-connects are already at 100G and soon 400G will be the norm. Yet as speeds increase, infrastructure managers will have to maintain that momentum while living within their power constraints.
A major challenge for data centres is to reduce power consumption across their infrastructure while delivering high-speed connectivity and feeding the growing demand for data. A study last year found that data centres globally had consumed well over 400 terawatt hours of electricity – far higher than the UK’s total consumption – and this could triple in the coming decade. As pressure mounts on data centres to reduce energy consumption, some ICPs have looked to colder climates such as the Nordics for facilities. Kolos, a US – Norwegian joint venture, is working on the world’s largest data centre in the Arctic Circle that could tap into hydropower and cut energy costs by 60%.
Tested to the limit
As ICPs continue to expand, they will build more data centres to accommodate the rising levels of content and require seamless DCI to deliver services to users at lightning fast speeds. Given the pace at which these businesses have grown, ICPs have had little time to put in place the rigorous procedures necessary for testing to ensure seamless DCI. This has been a major challenge for some data centre managers who have also had to grapple with the rising costs of cabling infrastructure as well as a plethora of protocols to interoperate.
All these challenges might seem like a tsunami, but there are steps that data centre managers can take in the realm of test and measurement – inside the data centre, within DCI and in network monitoring – to steady the ship.
Within the data centre, automated testing tools can inspect and certify fiber end faces for faster network build-outs and test functionality for MPOs (multi-fiber push-on). Effective AOC (Active Optical Cables) and DAC (Direct Attach Cable) test practices are essential to ensure optimum network performance and to address the challenges brought on by the growth of multifiber connectivity.
To stay ahead and prepare for increasing DCI speeds, ICP engineering labs need to test 400G interfaces with a versatile platform that can handle different applications and ports. Running simultaneous test modules, comparing and evaluating the results and performance of open APIs/protocols such as NETCONF/YANG on racks at high speeds of 100G, 200G and 400G can help to pinpoint potential issues and troubleshoot infrastructure complications before they arise.
Network monitoring needs to be automated and virtualized so that data centre managers have the capability to monitor, diagnose and resolve anomalies on virtual, physical and cloud-based infrastructure. The fibre networks they depend on require robust testing from end-to-end to maintain peak performance.
The data and content mind-set
Similar to how oil transformed economies in the 20th century, data is the world’s most valued resource today. The ICPs driving the new ‘data economy’ have experienced a meteoric rise. It is a similar tale with data traffic – it has grown in leaps and bounds. Data centre managers have had to adopt a new mind-set to stay ahead of new platforms, practices and protocols – not to mention data speeds – to keep pace with the ICP business model. Test and measurement is no different. To support data centre infrastructure and stay ahead of the curve for ICPs, test and measurement needs to be agile, virtual and automated.
