The 2007 launch of Windows Vista was such a disaster that Microsoft Research did a postmortem study of what went wrong, and the results set Microsoft up for its recent comeback. (Microsoft Photo)
Welcome to Mostly Cloudy, where you’ll find all the cloud tech news that trickled out during one of the slowest weeks of the year! In this edition, we’ll look at what Microsoft Research learned from the disaster that was Windows Vista, and how it applies to the modern cloud era.
This Week in Cloud: Clear direction, clear feedback, no bugs?
Software bugs are inevitable, but organizational dysfunction can make them worse. That was the conclusion from a decade-old study conducted by Microsoft Research that made the rounds this week thanks to a blog post from August Lilleaas, and it has a lot of relevance today as a generation of software developers shifts from older tools and techniques to the ones that make most sense for the cloud.
Windows Vista was a PC operating system released in late 2007, and it was the clearest sign as of that point that something was deeply, deeply wrong at Microsoft. It was delayed several times amid shifting requirements and technical challenges, years during which Apple unveiled its famous Mac vs. PC ads and Windows became something of a punchline.
As part of a postmortem into what went wrong, Microsoft Research posted a report in 2008 detailing how it examined “organizational metrics” like engineering turnover and structure alongside traditional software quality metrics in order to see what was having the most impact. “Our results provide empirical evidence that the organizational metrics are related to, and are effective predictors of failure-proneness,” researchers wrote in their abstract.
So what does an 11-year-old esoteric research have to do with cloud computing? Organizational structures are being upended as concepts like DevOps become mainstream, and teams that have years of experience building software for their own server equipment are being forced to learn new tricks to get the most out of their investment in cloud computing.
Lilleaas gets into a lot of technical and statistical analysis in his post, but here’s the gist: “At any rate, this discovery coupled with the findings from Accellerate leads me to at the very least believe that social elements is probably under-measured in software projects, and should be taken more seriously.”
Facebook’s Boston office. (Facebook Photo)
The flexibility of cloud computing is paid for by its complexity, and that’s one of many reasons why software teams should be thinking about finding ways to smooth out their workflows in the new year. It’s also incumbent on the non-technical leadership at these companies to avoid micromanaging the development process, avoiding classic points of frustration like last-minute feature requests and vague project guidelines.
Microsoft pulled itself out of a post-Vista tailspin by refocusing its energy on cloud computing, setting clear expectations over the last five years for the type of company it expects to be. It remains a complex organization, but for the first time in many years appears to be moving in a unified direction.
Best practices for operating on the cloud are already established, but they are becoming more evenly distributed as more and more companies modernize their tech infrastructure. Developers are going to have to have enough to worry about as they make this transition, and the easiest way for their companies to help them along this path is to structure teams, goals, and methods as simply as possible.
And Now, A Word…
Mostly Cloudy will be operating on a limited production schedule over the next few weeks, for obvious reasons. No matter what your holiday season entails, enjoy the rest before what is likely to be one of the craziest years most of us will ever see.
And thanks to the Mostly Cloudy subscribers who stepped up to provide funding for the Phoenix Counseling Center, which will be able to modernize its own approach to providing much-needed services to the Rogue Valley.
Around the Cloud
We're Reddit's Infrastructure team, ask us anything! (Reddit)
This was an interesting read, given Reddit’s status as one of the most popular sites on the internet and the traffic load that produces. The company runs almost everything on Amazon Web Services, but uses Google for a few services, and engineers shared a lot of detail about how they keep the site up and running.
A Rare Russian Tech Triumph, a Police Raid and a Backlash (The New York Times)
Nginx’s services are a core part of the web, and it looks like sanity has prevailed in Russia, where the company’s offices were raided at gunpoint last week as part of a dispute over the ownership of those of that code. Rambler, the partially state-owned company that alleged Ngnix executives stole its code while working for Rambler before Ngnix was founded, said it didn’t want Russian authorities to pursue the case after facing a backlash from the local business community, fearful the incident would harm investment in the region.
Kubernetes is booming, but consolidation is coming (ZDNet)
This is a nice overview of the state of Kubernetes a few weeks after the 2019 KubeCon conference in San Diego. The much-discussed container orchestration project is evolving into both a well-known platform for cloud-native software development and a less visible part of the stack, as new abstractions start to emerge.
The INQUIRER reaches end-of-life (The Inquirer)
Back in the day, The Inquirer (and its sibling/rival The Register) were rare bright spots of wit and wisdom in the drab world of enterprise tech coverage, driven mostly by U.S. trade magazine editors. I learned so much about processors, Intel, and AMD from the Inquirer, and it’s sad to witness the demise of yet another tech media company.
Can the Internet Survive Climate Change? (The New Republic)
A solar farm for a Google data center in Europe. (Google Photo)
We’re talked about the climate impact of cloud computing and software development before, and this report makes a very good point: more than half the world has yet to connect to the internet on a regular basis. As they arrive, demands on the system will grow, which is why some researchers are experimenting with alternative network designs and fuel sources to prepare.
Amazon Conference Badges Tracked Attendees' Movements (Vice)
Earlier this month at re:Invent I noticed a weird plastic cube-like thing with a QR-code on my badge lanyard but didn’t really think much of it, until I saw this report from Vice. The sea of humanity that invades the Venetian for re:Invent every year could stand to be better managed, for sure, but it appears like AWS buried the disclosure and opt-out option for this “feature” (I definitely missed it).
Is VMware Winning the Cloud Wars? (SDxCentral)
Can we please stop with the “wars” metaphors in headlines? Nobody has benefited more from the emergence of the hybrid cloud operating model than VMware, which bought itself time to adjust during a period in which many people thought it was doomed and has since invested in cloud-native expertise like Pivotal and Heptio.
NOAA Contracts With 3 Major Cloud Providers on Big Data Project (NextGov)
That being said, I’m not sure how anyone could resist making a pun in the headline for a story about cloud computing and weather research. This deal will allow NOAA, which collects a massive amount of weather data around the country, to make it more widely available to researchers and the public under a multiyear deal with AWS, Microsoft, and Google.
Randy Suess, Computer Bulletin Board Inventor, Dies at 74 (The New York Times)
Bulletin boards were the first social internet product that a generation of people encountered, and while they were prone to a lot of the same problems as the modern social media world, they provided an essential building block for the web as we know it today. Raise a glass to Randy Suess and his family over the holiday break.
The AWS Cloud Extends to the Edge, With Likely Boost for Colo Providers (Data Center Frontier)
AWS’s edge computing strategy was one of the major topics of discussion at re:Invent, and this post takes a closer look at how it much play out at the “edge,” which is really just the existing collection of managed data centers and telecom provider equipment that sit far from AWS’s data centers. One interesting note: lots of smaller companies have been defining standards and practices around edge computing as larger players tiptoe in, and that could be about to change.
Facebook is entering the cloud-gaming market with its PlayGiga acquisition (Business Insider)
The discussion around Facebook’s possible cloud-services pivot just got a little more interesting! Somebody is going to figure out how to deliver top-tier games to demanding gamers over the internet, and there’s no reason why Facebook and its big advertising department couldn’t pull it off.