Facebook improves connectivity in the African region with an undersea internet cable

According to the Uptime Institute's Data Center Industry Survey 2021, there is still a shortage of skilled workers in the industry, and artificial intelligence is not expected to help solve this problem in the near future. As the demand for digital skills grows, the scarcity is becoming more acute. In addition, today's data center teams increasingly require both operational and development skills. The data center engineer must still be involved in preventive maintenance of equipment, but also have the technical knowledge to use mobile applications and be able to dig into any type of system and extract telemetry from it. Now everything can be viewed remotely, listened to and touched with the help of software - and this can be done even before the engineer enters the room. This goes beyond the traditional understanding of the structure of data centers and the use of software. The rise in automation will foster the use of AI and machine learning applications and tools, as well as tools capable of analyzing large amounts of data. This opens up opportunities for people who can work as analysts and data scientists who are able to use the latest tools to extract insights from massive amounts of data. What new job responsibilities should today's data center professionals prepare for? What kind of education do they need to get or what skills do they need to master? Data center professionals can handle the transition to new roles in several ways. First, they should be familiar with the underlying architecture of the various public and private clouds. Second, they must hone their skills in critical areas such as infrastructure as code, DevOps, intelligent analytics, data operations, container assembly and operation (Kubernetes), service meshes, and serverless technologies. With this knowledge, they will be well prepared to manage data centers now and in the future. In many ways, finding a unicorn now means finding someone who combines diagnostic thinking with the ability to work on a technology platform. I don’t think we’ll ever replace a data center engineer, but we will need to combine the two extremes - integrating software and training existing technicians who have attended vocational schools or are certified in electrical engineering and control, ventilation / mechanics and telecommunications / cable structure. It's about the mindset of a mechanic with data overlay.

Data from the Uptime Institute's annual survey of data center disruptions and their causes shows fewer serious disruptions. Only 6 percent of respondents reported serious incidents (“Category 5”) in 2020. A year earlier, such accidents were reported by 11 percent of those surveyed. Interestingly, against the background of a decrease in the number of major accidents in data centers, financial and reputational damage from such incidents, unfortunately, is increasing, which, according to experts, is due to the continuously increasing dependence of business and government organizations on IT infrastructure. In particular, when commenting on recent significant accidents, more than half of the respondents who reported downtime in the past three years estimated the cost of accidents at more than $ 100,000 (with about a third claiming damages of $ 1 million or more). UPS circuit breakers and batteries should be carefully checked The Uptime Institute used survey data in conjunction with direct analysis of customer power use patterns to draw some conclusions about the causes of downtime. Local power outages continue to be the most common cause of power-related data center outages. However, the components of these systems, such as UPS batteries and circuit breakers, are often the most vulnerable points of failure. The organization's experts noted that in order to reduce the cost of building a data center, owners and operators of facilities are advised to install so-called Distributed Redundant Systems (DRS). The implementation of this concept in practice involves the deployment of two independent arrays of batteries that form the UPS, each of which is able to take on the entire load, and not just part of it. Human factor While most of the blame for data center downtime lies with technology, the notorious human factor also needs to be considered. A recent survey by Uptime Institute has shown that accidents caused by data center operators are quite common. In the survey, 42% of respondents said they have had a failure due to human error in the past three years. Among those surveyed, 57% cited improper work of data center personnel (non-compliance with procedures) as the cause of the accident, and the remaining 43% cited incorrect processes / procedures as the main reason for downtime. The study clearly shows that an increased emphasis on effective human resource management and continuous training of data center personnel will lead to better service efficiency for critical infrastructure and minimize the risk of downtime.

According to analysts' forecasts, tens of billions of IoT devices will work in the world in the near future. The old IPv4-based addressing system, which was more or less capable of providing IP addresses to the “Internet of people”, is practically useless in the era of the coming “Internet of machines”. A new protocol, IPv6, aims to solve the problem. What does IPv6 give? The documents defining the new Internet protocol were issued by the Internet Engineering Task Force back in the mid-90s, and the official launch of the IPv6 protocol on an ongoing basis took place on June 6, 2012. Many companies started switching to it earlier, for example Google - since 2008. The protocol got the number "6" because the name IPv5 was reserved for an experimental real-time protocol that never came out. But it also did not disappear at all - many of the concepts embodied in it can be found in the MLPS protocol. Thanks to the 128-bit addressing scheme embedded in IPv6, the number of network addresses available in it is 2 to the power of 128. Such a large address space makes it unnecessary to use NAT (there are enough addresses for everyone) and simplifies the routing of data. For example, routers no longer have to fragment packets, it is now possible to forward large packets up to 4 GB in size. Implementation of IPv6. What is holding him back? 8 years after the official launch, the IPv6 protocol is gradually being implemented in the networks of telecom operators, as well as Internet service providers in different countries, coexisting with its predecessor - the IPv4 protocol. The most active use of the new addressing system is mobile operators and Internet providers. For example, according to the industry group World IPv6 Launch, T-Mobile USA uses IPv6 for almost 95% of the traffic volume, and Sprint Wireless - 89%. There are fans of progress in other countries - Indian Reliance Jio Infocomm (90%), Brazilian Claro Brasil (66%). Of the Russian operators, MTS is the highest in the ranking, at 83rd place, with 55%. In the country rankings of IPv6 penetration, according to Google, the leaders are Belgium (52.3%), Germany (50%), India (47.8%), Greece (47.6%). The United States has only 40.7%, less than, for example, Vietnam (43.1%). Russia has nothing to brag about (5.6%). However, China has 0.34% at all. When will IPv4 be "disabled"? Formally, no one will most likely disable IPv4. When the new protocol was launched in June 2012, the Dual Stack model was chosen, in which IPv4 and IPv6 networks operate in parallel. In most of the world, new IPv4 addresses "ran out" between 2011 and 2018, but today it is clear that such network addresses will be sold and reused for quite some time. However, as the number of transitions to the new protocol grows, operators and ISPs can be expected to start charging companies for IPv4 addresses while IPv6 becomes free of charge. Perhaps some services will become available only in the next generation networks. IoT networks will operate on IPv6. And, with the natural obsolescence of IPv4 equipment (and little new will be released), the world will gradually move away from IPv4.

Due to the Covid-19 pandemic, interruptions in the operation of broadband Internet and data centers are especially acute for ordinary users, government agencies and businesses. High-speed internet has been a lockdown lifesaver for millions of office workers, students, healthcare professionals and others who are increasingly using broadband for telecommuting / learning. In Oakland, a data center shutdown occurred due to rodents that damaged the cable At the end of January 2021, over 1,000 broadband Internet users in the western part of the New Zealand city of Auckland were disconnected from the network for more than a day. An investigation found that the internet outage was due to rodents damaging a 144-wire fiber optic cable belonging to the telecommunications company Chorus. The incident took place in the Massey area. According to Chorus, the cable break "was probably due to a rodent, which is related to similar damage seen elsewhere." Pakistani data centers go offline due to malfunction of submarine cable A similar incident happened in early February 2021 in Pakistan. Much of the local Internet infrastructure was interrupted by a damaged submarine cable near the Abu Talata settlement in Egypt. The corresponding trunk cable is used by Pakistani ISP Trans World Associates (TWA), which serves about 40 percent of the country. The exact information about the reasons for the incident was not disclosed. But the previously considered Asian country has already faced problems of a similar nature, when backbone fiber optic cables were inadvertently damaged and even deliberately cut. Interestingly, in January 2021, Pakistani data center operators faced a power outage that swept across the country. As a result, most of the Internet services served by local data centers went offline. Downtime of the AWS data center has led to numerous clients going offline, including Roku, Adobe On November 25, 2020, the Amazon Kinesis real-time streaming service ceased operations at the AWS data center campus in Northern Virginia. Kinesis is being used by other AWS tools that have also stopped working. This circumstance led to the shutdown of the services of a number of clients, including Flickr, iRobot, Roku. The system was fully operational by the evening of the same day. The company apologized for the problems, noting how important this service is to customers and their customers. Massive shutdown of server farms in Texas due to storm In the second half of February 2021, due to data center outages caused by a power outage in Texas, North American Internet users faced many popular services going offline. The state has put in place a plan to phase out power due to storm Uri. A poorly funded and poorly regulated regional power grid has lost tens of gigawatts of electricity, leaving millions of citizens without heating. There are about 2 hundred data centers operating in Texas. Most of the major local data center operators switched to diesel generators when faced with reduced power flow from the central network, while continuing to serve corporate customers and ordinary users.