 |
|
|
|||||
When internet services platform Cloudflare suffered an outage in November, it took a big chunk of the online world down with it. Major platforms like ChatGPT, X, and Canva became unreachable. So did digital services offered by countless banks, retailers, and many other businesses. During the six-hour meltdown, as many as 2.4 billion users could have felt the impact. Software outages like this have always been and always will be part of online life. But today our systems are more interconnected than ever, so a single failure can ripple outward. AI only amplifies that risk. Yet, too many companies still lack protection against such disasters. In an era when outages are inevitable, theyre effectively operating without a safety net. The fundamental missing ingredient is something simple but easily overlooked: resilience testing. In a nutshell, resilience testing is all about pressure testing your software, before issues happen. It ensures that systems keep workingor quickly bounce backwhen things go wrong. Think of resilience testing as a small safety step to prevent big problems. The annual median cost of a high-impact IT outage is about $76 million. Businesses can also suffer reputational damage, lose customers, and get hit with regulatory penalties. Cloudflare is only one recent example. In the past year alone, AWS, Microsoft 365, and Starlink all went down, to name just a few. So why arent more businesses stress-testing their software for inevitable failure? Heres why, and what companies can do about it. MOST COMPANIES DONT BOTHER WITH RESILIENCE TESTING As high as the stakes are, businesses have reasons to avoid software resilience testing. The process is technical, and it can get messy. Modern resilience testing, also called chaos engineering, was put in the spotlight 15 years ago by Netflix software developers. Realizing that the only way to test for resilience is to simulate problems in the wild or in production, they created a suite of tools that replicated network crashes, cloud services meltdowns, and other real-world failures. Netflix might have been able to roll with the punches, but few other companies have the expertise or the stomach to compromise their systems like this. Its the equivalent of starting a controlled fire to ensure you have the resources to put it out. Resilience testing requires the technical acumen to know what failures to simulate for and responses to take. Putting these drills into action also entails risk, like triggering your homes fire sprinkler system which could ruin the furniture. Most importantly, developers need to know what to do when tests reveal weaknesses. Because the threshold for resilience testing is so high, it isnt integrated into most companies software development processes. Theres rarely a dedicated team, and often no one except maybe the CTO is clearly in charge. As a result, resilience testing becomes a bottleneck, so companies dont bother with it. A BETTER WAY FORWARD: HELP FROM AI The good news: It no longer has to be this way. For companies that want to adopt resilience testing, new platforms and toolspowered by AIare making the process safer and easier. Specialized resilience testing agents now enable companies to automate and optimize testing, without needing dedicated experts or teams. First, the AI agent identifies likely edge casesunusual or unexpected scenarios that could compromise reliability. It examines system behavior in production, how services interact, and where similar systems have previously failed. For example, the agent might highlight a scenario where a service slows, rather than fails outright. Another edge case: A code deployment updates only half the companys servers, leading to inconsistent user experiences. The agent then generates and prioritizes the test cases most likely to reveal resilience issues, explaining why each one matters. It can also set up and run those tests. After problems are identified, the AI agent suggests targeted fixes, making the software more resilient. With the heavy lifting completed, developers can review and apply those insights. WHY RESILIENCE TESTING NEEDS TO SHIFT LEFT Having the right tools is one thing, but effective resilience testing requires more than just software. Creating a culture of resilience is part of the solution. Software teams need to include testing in their routine. Ultimately, the only way to strengthen yourself against failures is to practice for them. If you never run those drills, you never know how bad things can get until its too late. Developers should also remember that resilience testing isnt just about full-scale, five-alarm outages. Its also about small, partial failures that create a poor user experience for customers, without necessarily taking the whole system down. Lets say a platform like Cloudflare has an issue affecting a major banks consumer app, leaving millions unable to check their balances. Resilience testing should anticipate this problem and provide a viable workaround. But the best way to encourage a culture of resilience is to shift leftmoving resilience testing to the software development preproduction phase, before code ever goes live. Shifting left helps teams catch weaknesses long before customers feel them. Thats crucial with todays complex, interconnected software systems, where seemingly minor issues can rapidly spiral into major outages. Rather than scramble to diagnose problems during live incidents, developers can uncover and fix them in a safe environment. Shifting left can save money and stress, too. Fixing resilience issues in production is costly and disruptive, often pulling team members away from other vital tasks. By taking a proactive approach, developers and business leaders can be more confident in the product they deliver to customers. Ultimately, resilience testing isnt rocket science. Companies that run fire drills for their software and embrace a culture of resilience testing will find themselves in a stronger position when the next disruption strikes. And in an increasingly interconnected world, where AI tools and features depend on more underlying services than ever, its safe to say that might be sooner rather than later. Jyoti Bansal is CEO of Harness.
Category:
E-Commerce
If you’re a millennial or simply a fan of sci-fi, you likely remember 2010’s smash hit Inception, written and directed by Christopher Nolan. The story follows Dom Cobb (played by Leonardo DiCaprio), a professional thief who specializes in stealing privileged secrets from peoples minds while they dream. He uses advanced technology to enter another persons subconscious while theyre asleep and take whatever information he wants. The problem DiCaprio’s character faces is that the line between reality and dreams become blurred, making it increasingly difficult to tell whats real. In many ways, especially for retail investors, that’s a picture of the modern market. A place where rumor, rhetoric, and anticipation can impact price action in unpredictable ways. When you add the ever-changing realities of our current political and macro-economic environment, the average Joe trader can feel like the odds are stacked against him. But there is one set of data that more often than not, cuts through the noise. And that’s the options market. OPTIONS PROVIDE ACCURATE VIEWS Options give us a more accurate picture of what the people who move the markets are actually thinking. Institutions and hedge funds make their living on positive returns. And while stocks can often be manipulated, options have a shorter shelf-life and are way less forgiving. And it’s there, in the nuances of the “put to call ratios” of the options market, that AI is quietly giving retail traders a helping hand. Let me give you an example. OnOctober 9th, our Prospero.ai AI-driven Net Options Sentiment (a gauge of real-time, market options data and positioning), plunged from a bullish 33 to an ominously bearish zero in a matter of minutes. No headline. No Fed speech. No earnings shock. Just a massive bearish shift, seemingly out of nowhere. We saw the shift and sent an alert to our followers on X. At 10:30 a.m. the next morning, the numbers were still sitting at zero. We issued another alert warning that a possible market reversal might be brewing. Thirty minutes later, without any warning, President Trump posted on social media about a looming trade war with China. The market fell off a cliff. Several fintech influencers noticed our early warnings and brought positive attention to it on X. Some were skeptical. @cyberdaddy mused, “Today’s selloff was 100% correlated to the China tariff threats, which were impossible for any person or algo to forsee.this was 100% random luck. But the reality is that this wasnt the first time our signals picked up on a major market move before it happened. The same pattern appeared in February when we saw SPY Net Options Sentiment lead the market down. We warned about it (Dont Trim the Hedges) on March 3rd, just a few days before a historic market correction on Liberation Day. Twice now, the data revealed massive bearish options positioning before a drop in price. WHATS REALLY HAPPENING? That brings us to an important question. If the data is picking up on signals before theyre public, whats really happening? Is AI becoming smart enough to sense fear before humans can express it? Or is information, intentional or not, pulsing through back channels in a way once thought unimaginable 20 years ago? In earlier eras, Wall Street’s most valuable information moved in whispers and was confined to smoke-filled rooms and the back nine of golf courses. Today, that same information moves with lightning speed through social media threads, encrypted group chats, and algorithmic pattern recognition. One whisper in the right chat room can ripple through a thousand terminals before the SEC finishes its coffee. A policymakers assistant, a janitor overhearing a senator’s conversation as he informs a colleague of a military contract, or heightened tensions in the White House can all lead to that same outcome. The lines between what’s public, private, and privileged are dissolving. So maybe the question isnt whether AI is getting smarter or if those in the know are getting bolder. Maybe it’s that AI is hearing the market’s whispers more clearly than ever before. Cyberdaddy backed off his protest of the validity of our system: After doing some more research, I found that the de-risking and drop in options sentiment likely occurred for logical reasons that your model picked up on. The China export-control announcement on October 8th probably drove much of that defensive positioning among people paying attention to geopolitics (which major firms obviously do). My apologies for jumping the gun, I was wrong. THE CRUX OF THE ISSUE This is the crux of the matter. Did a shift in the perception of risk regarding the situation with China lead to a massive hedge or bet against the QQQ (an index including 100 of the largest non-financial companies listed on the Nasdaq exchange, heavily weighted toward technology and growth)? Or were some more in the know about the political winds changing? How could we even begin to differentiate between the two? By the way, the word inceptionisn’t just a movie title, it’s a word that means the beginning, or start of something brand new. And maybe we are at a point of inception. Because for the first time, AI isn’t just leveling the playing field; it’s changing who gets to play on that field in the first place. George Kailas is CEO of Prospero.ai.
Category:
E-Commerce
A wild winter storm was expected to bring strong winds, heavy snow and frigid temperatures to the Great Lakes and Northeast on Tuesday, a day after a bomb cyclone barreled across the northern U.S. and left tens thousands of customers without power.The storm that hit parts of the Plains and Great Lakes on Monday brought sharply colder air, strong winds and a mix of snow, ice and rain that led to treacherous travel. Forecasters said it intensified quickly enough to meet the criteria of a bomb cyclone, a system that strengthens rapidly as pressure drops.Nationwide, more than 153,000 customers were without power early Tuesday, more than a third of them in Michigan, according to Poweroutage.us.As Monday’s storm moved into Canada, the National Weather Service predicted more inclement weather conditions for the Eastern U.S, including quick bursts of heavy snow and gusty winds known as snow squalls.New York Gov. Kathy Hochul warned that whiteout conditions were expected Tuesday in parts of the state, including the Syracuse-metro area.“If you’re in an impacted area, please avoid all unnecessary travel,” she said in a post on the social media platform X.Snow piled up quickly in Michigan’s Upper Peninsula on Monday, where as much as 2 feet (60 centimeters) fell in some areas, according to the National Weather Service. Meteorologist Ryan Metzger said additional snow was expected in the coming days, although totals would be far lighter.Waves on Lake Superior that were expected to reach 20 feet (6 meters) Monday sent all but one cargo ship into harbors for shelter, according to MarineTraffic.com. Weather forecasting on the lakes has improved greatly since the Edmund Fitzgerald sank in 1975 after waves were predicted at up to 16 feet (4.8 meters).The fierce winds on Lake Erie sent water surging toward the basin’s eastern end near Buffalo, New York, while lowering water on the western side in Michigan to expose normally submerged lakebed even the wreck of a car and a snowmobile.Kevin Aldrich, 33, a maintenance worker from Monroe, Michigan, said he has never seen the lake recede so much and was surprised on Monday to spot the remnants of old piers dating back to the 1830s. He posted photos on social media of wooden pilings sticking up several feet from the muck.“Where those are at would typically be probably 12 feet deep,” he said. “We can usually drive our boat over them.”Dangerous wind chills plunged as low as minus 30 F (minus 34 C) across parts of North Dakota and Minnesota on Monday. And in northeast West Virginia, rare, nearly hurricane-force winds were recorded on a mountain near Dolly Sods, according to the National Weather Service.In Iowa, after blizzard conditions eased by Monday morning, high winds continued blowing fallen snow across roadways, keeping more than 200 miles (320 kilometers) of Interstate 35 closed. State troopers reported dozens of crashes during the storm, including one that killed a person.On the West Coast, the National Weather Service warned that moderate to strong Santa Ana winds were expected in parts of Southern California through Tuesday, raising concerns about downed trees in areas where soils have been saturated by recent storms. Two more storms were forecast later this week, with rain on New Year’s Day potentially soaking the Rose Parade in Pasadena for the first time in about two decades. Associated Press writers Julie Walker in New York; Corey Williams in Detroit; Margery Beck in Omaha, Nebraska; Susan Haigh in Norwich, Connecticut; and Becky Bohrer in Juneau, Alaska, contributed. Leah Willingham and Jeff Martin, Associated Press
Category:
E-Commerce
All news |
||||||||||||||||||
|
||||||||||||||||||