In practice, to hold teams accountable for what they develop, processes need to shift left to earlier in the development lifecycle, where development teams are. By moving steps like testing, including security testing, from a final gate at deployment time to an earlier step, fewer mistakes are made, and developers can move more quickly. The principles of shifting left also apply to security, not only to operations. It’s critical to prevent breaches before they can affect users, and to move quickly to address newly discovered security vulnerabilities and fix them. Instead of security acting as a gate, integrating it into every step of the development lifecycle allows your development team to catch issues earlier. A developer-centric approach means they can stay in context and respond to issues as they code, not days later at deployment, or months later from a penetration test report. Shifting left is a process change, but it isn’t a single control or specific tool—it’s about making all of security more developer-centric, and giving developers security feedback where they are. In practice, developers work with code and in Git, so as a result, we’re seeing more security controls being applied in Git.
As your system grows, the connections between microservices become more complex. Communicating in a fault-tolerant way, and keeping the data that is moving between services consistent and fresh becomes a huge challenge. Sometimes microservices must communicate in a synchronous way. However, using synchronous communications, like REST, across the entire deep system makes the various components in the chain very tightly coupled to each other. It creates an increased dependency on the network’s reliability. Also, every microservice in the chain needs to be fully available to avoid data inconsistency, or worse, system outage if one of the links in a microservices chain is down. In reality, we found that such a deep system behaves more like a monolith, or more precisely a distributed monolith, which prevents the full benefits of microservices from being enjoyed. Using an asynchronous, event-driven architecture enables your microservices to publish fresh data updates to other microservices. Unlike synchronous communication, adding more subscribers to the data is easy and will not hammer the publisher service with more traffic.
"The jobs aren't the same as two or three years ago," he acknowledges. "The types of skill sets employers are looking for is evolving rapidly." Three factors have led the evolution, O'Malley says. The first, of course, is COVID-19 and the sudden need for large-scale remote workforces. "Through this we are seeing a need for people who understand zero-trust work environments," he says. "Job titles around knowing VPN [technology] and how to enable remote work with the understanding that everyone should be considered an outsider [are gaining popularity]." The next trend is cloud computing. With more organizations putting their workloads in public and private clouds, they've become less interested in hardware expertise and want people who understand the tech's complex IT infrastructure. A bigger focus on business resiliency is the third major trend. The know-how needed here emphasizes technologies that make a network more intelligent and enable it to learn how to protect itself. Think: automation, artificial intelligence, and machine learning. The Edge asked around about which titles and skills security hiring managers are interested in today.
The Agile Manifesto prioritizes working software over comprehensive documentation -- though don't ignore the latter completely. This is an Agile FAQ for newcomers and experienced practitioners alike, as many people mistakenly think they should avoid comprehensive documentation in Agile. The Agile team should produce software documentation. Project managers and teams should determine what kind of documentation will deliver the most value. Product documentation, for example, helps customers understand, use and troubleshoot the product. Process documentation represents all of the information about planning, development and release. Similarly, Agile requirements are difficult to gather, as they change frequently, but they're still valuable. Rather than set firm requirements at the start of a project, developers change requirements during a project to best suit customer wishes and needs. Agile teams iterate regularly, and they should likewise adapt requirements accordingly. ... When developers start a new project, it can be hard to estimate how long each piece of the project will take. Agile teams can typically gauge how complex or difficult a requirement will be to fulfill, relative to the other requirements.
This might seem a strange piece of research for Facebook to focus on. Better news feed algorithms? Sure. New ways of suggesting brands or content you could be interested in interacting with? Certainly. But turning 2D images into 3D ones? This doesn’t immediately seem like the kind of research you’d expect a social media giant to be investing. But it is — even if there’s no immediate plan to turn this into a user-facing feature on Facebook. For the past seven years, Facebook has been working to establish itself as a leading presence in the field of artificial intelligence. In 2013, Yann LeCun, one of the world’s foremost authorities on deep learning, took a job at Facebook to do A.I. on a scale that would be almost impossible in 99% of the world’s A.I. labs. Since then, Facebook has expanded its A.I. division — called FAIR (Facebook A.I. Research) — all over the world. Today, it dedicates 300 full-time engineers and scientists to the goal of coming up with the cool artificial intelligence tech of the future. It has FAIR offices in Seattle, Pittsburgh, Menlo Park, New York, Montreal, Boston, Paris, London, and Tel Aviv, Israel — all staffed by some of the top researchers in the field.
The companies that understand the potential impact of quantum computing on their industries, are already looking at what it would take to introduce this new computing capability into their existing processes and what they need to adjust or develop from scratch, according to Uttley. These companies will be ready for the shift from “emergent” to “classically impractical” which is going to be “a binary moment,” and they will be able “to take advantage of it immediately.” The last stage of the quantum evolution will be classically impossible—"you couldn’t in the timeframe of the universe do this computation on a classical best-performing supercomputer that you can on a quantum computer,” says Uttley. He mentions quantum chemistry, machine learning, optimization challenges (warehouse routing, aircraft maintenance) as applications that will benefit from quantum computing. But “what shows the most promise right now are hybrid [resources]—“you do just one thing, very efficiently, on a quantum computer,” and run the other parts of the algorithm or calculation on a classical computer. Uttley predicts that “for the foreseeable future we will see co-processing,” combining the power of today’s computers with the power of emerging quantum computing solutions.
Data preparation for ML is deceptive because the process is conceptually easy. However, there are many steps, and each step is much more complicated than you might expect if you're new to ML. This article explains the eighth and ninth steps ... Other Data Science Lab articles explain the other seven steps. The data preparation series of articles can be found here. The tasks ... are usually not followed in a strictly sequential order. You often have to backtrack and jump around to different tasks. But it's a good idea to follow the steps shown in order as much as possible. For example, it's better to normalize data before encoding because encoding generates many additional numeric columns which makes it a bit more complicated to normalize the original numeric data. ... A complete explanation of the many different types of data encoding would literally require an entire book on the subject. But there are a few encoding techniques that are used in the majority of ML problem scenarios. Understanding these few key techniques will allow you to understand the less-common techniques if you encounter them. In most situations, predictor variables that have three or more possible values are encoded using one-hot encoding, also called 1-of-N or 1-of-C encoding.
NIST notes that zero trust is not a stand-alone architecture that can be implemented all at once. Instead, it's an evolving concept that cuts across all aspects of IT. "Zero trust is the term for an evolving set of cybersecurity paradigms that move defenses from static, network-based perimeters to focus on users, assets and resources," according to the guidelines document. "Transitioning to [zero trust architecture] is a journey concerning how an organization evaluates risk in its mission and cannot simply be accomplished with a wholesale replacement of technology." Rose notes that to implement zero trust, organizations need to delve deeper into workflows and ask such questions as: How are systems used? Who can access them? Why are they accessing them? Under what circumstances are they accessing them? "You're building a security architecture and a set of policies by bringing in more sources of information about how to design those policies. ... It's a more holistic approach to security," Rose says. Because the zero trust concept is relatively new, NIST is not offering a list of best practices, Rose says. Organizations that want to adopt this concept should start with a risk-based analysis, he stresses.
Early threat detection and response is clearly part of the answer to protecting increasingly connected networks, because without threat, the risk, even to a vulnerable network, is low. However, ensuring the network is not vulnerable to adversaries in the first place is the assurance that many SOCs are striving for. Indeed, one cannot achieve the highest level of security without the other. Even with increased capacity in your SOC to review cyber security practices and carry out regular audits, the amount of information garnered and its accuracy, is still at risk of being far too overwhelming for most teams to cope with. For many organisations the answers lie in accurate audit automation and the powerful analysis of aggregated diagnostics data. This enables frequent enterprise-wide auditing to be carried out without the need for skilled network assessors to be undertaking repetitive, time consuming tasks which are prone to error. Instead, accurate detection and diagnostics data can be analysed via a SIEM or SOAR dashboard, which allows assessors to group, classify and prioritise vulnerabilities for fixes which can be implemented by a skilled professional, or automatically via a playbook.
GDPR fines are like buses: You wait ages for one and then two show up at the same time. Just days after a record fine for British Airways, the ICO issued a second massive fine over a data breach. Marriott International was fined £99 million [~$124 million] after payment information, names, addresses, phone numbers, email addresses and passport numbers of up to 500 million customers were compromised. The source of the breach was Marriott's Starwood subsidiary; attackers were thought to be on the Starwood network for up to four years and some three after it was bought by Marriott in 2015. According to the ICO’s statement, Marriott “failed to undertake sufficient due diligence when it bought Starwood and should also have done more to secure its systems.” Marriott CEO Arne Sorenson said the company was “disappointed” with the fine and plans to contest the penalty. The hotel chain was also fined 1.5 million Lira (~$265,000) by the Turkish data protection authority — not under the GDPR legislation — for the beach, highlighting how one breach can result in multiple fines globally.
Quote for the day:
"Making the development of people an equal partner with performance is a decision you make." -- Ken Blanchard