Welcome to Mostly Cloudy, and Happy New Year! Today: how Southwest did more in one weekend for companies hawking digital transformation than they did for themselves all year, why this could be the year of the “r” word, and Twitter is down to two data centers.
Photo: Southwest
You are now free to move about the data center
Mainstream television ads for enterprise tech companies always have a certain level of disconnect from the actual grind of keeping a digital business up and running in the 21st century, pitched as they are to finance executives and retired investors. Those ads talk about data, streamlining operations, and finding efficiencies, painting a gauzy picture of how “digital transformation” will make your company richer and better-looking.
Professionals find it easy to dismiss such fluff, but there is an underlying truth behind those ads: You can get away with deferring maintainance and upgrades to the crucial software systems that run key parts of a business for a certain period of time, but the longer you wait, the more likely that software is to blow up in your face and the more difficult it will be to recover. You’d think an airline would understand the risks of deferred maintenance, but here we are.
Southwest’s epic meltdown over the Christmas weekend, which disrupted its operations for more than a week, appeared to be the result of years of deferred upgrades to homegrown software systems that looked “historic like they were designed on Windows 95,” the head of the Southwest flight attendants union told the Dallas Morning News.
Employees had been complaining about the outdated systems for years, according to the New York Times. Widespread severe weather and Southwest’s unique point-to-point strategy contributed to greatly to its woes, but its internal crew-planning software required pilots and flight attendants to wait hours on hold on the freaking telephone simply to let the company know where they were, among other indignities.
The result was one of the most public corporate operational meltdowns in recent memory, in one weekend destroying an incalcuable amount of brand value that Southwest has built up over more than 50 years. It’s a wake-up call to any IT and finance departments returning from the holiday break knowing they’ve been putting off those updates.
There are many ways to go about modernizing outdated systems, and they don’t necessarily involve plunging into a shotgun marriage with a cloud vendor (more on that in a bit). The process is harder by definition for older companies like Southwest that invested huge sums of money building out their digital strategy 15 to 20 years ago, but nobody wants to be the next cautionary tale of allowing cost-minimization considerations to dictate tech investment strategies.
Happy New Year to enterprise tech sales reps: Southwest just gave you a fresh set of talking points.
We have to go back
“Repatration” replaced “multicloud” as the dirty word few AWS executives wanted to discuss last month in Las Vegas at re:Invent. Kicked off almost two years ago by a provocative blog post from a16z’s Sarah Wang and Martin Casado, the concept of cloud repatriation — the notion that startups built on the cloud will eventually find it necessary to manage their own data centers as cloud costs spiral — is quite inconvenient for the cloud leader.
But Deloitte’s David Linthicum, as prominent a cloud evangelist as there has been over the last decade, is starting to see signs that Wang and Casado were on to something. “Business is business.” Linthicum wrote in Infoworld Tuesday. “You can’t ignore the fact that it makes economic sense to move some workloads back to a traditional data center.”
A lot of separate issues collide when talking about cloud costs. As Linthicum points out, companies that “lift and shift” applications built for on-premises data centers into the cloud often find those applications need significant modifications to run efficiently on cloud servers, which takes effort. The complexity inherent in modern cloud services also makes it hard for some companies to keep track of everything they’re running in the cloud, where the meter is always on.
It’s hard to imagine the repatriation movement making a significant dent in the fortunes of AWS, Microsoft Azure, and Google Cloud just yet. But as the talent and skill that allowed Silicon Valley tech companies to domintate enterprise tech strategy becomes more evenly distributed across non-tech businesses, one of the key selling points of the cloud — we can run this stuff better than you can — starts to look dated.
Around the enterprise
Twitter closed its Sacramento data center on Christmas Eve, leaving the company with two main self-managed facilities as it frantically tries to cut costs to service its debt obligations.
Several European countries are looking at using the residual heat from Big Cloud data centers to keep residential properties warm.
What exactly will U.S. taxpayers get from the chip-manufacturing subsidy bonanza of 2021? A fair amount of construction projects that won’t reduce the world’s dependency on Taiwan’s chip factories any time soon, according to The New York Times.
The Register took a deep look at Apache Iceberg, the open-source data analytics project that is a foundational layer for Snowflake and Google, and a competitor of sorts to Databricks.
Thanks for reading — see you later this week!