AWS user’s data returned ‘because one human being inside AWS decided to give a damn’

A software engineer and developer, who previously accused Amazon Web Services (AWS) of a “digital execution,” has shared the good news that his data has now been restored. What made this impossible task possible was “one human being inside AWS [who] decided to give a damn,” according to data deficit disaster victim Abdelkader Boudih. The same insider source shared some interesting but worrisome insights about what went on behind the scenes during the days when robotic AWS reps insisted Boudih’s precious 10 years of data had been “terminated.”

In the earlier episode, we heard about an unnamed insider at AWS reaching out to Boudih. That character was never verified, though they apparently knew too much about this particular case to be a common prankster. Around the same time as we reported on the original debacle, an AWS employee had reached out via official channels to give hope that there would be some way to recover the precious data.

Tarus Balog was basically the first ‘human’ level contact from AWS, according to Boudih. Previous AWS-sourced responses were, in comparison, of a very scripted nature. However, Balog showed empathy and seemed to speak as someone with authority, telling the already long-suffering software engineer that leaders at AWS were aware of his data deletion blog, and were looking to make sure such cases were prevented, going forward.

Balog initially refrained from promising to recover the lost data, but cogs were definitely whirring at AWS at this point (August 5). The case was escalated to the top-level severity ticket available to ‘mortals’, and apparently the AWS CEO actually became aware of this particular incident at this time.

Account restored but tainted by 'deception'

The next morning (August 6), Boudih woke to the news – officially from Amazon – that his account had been restored. Though this was extremely welcome news, it was tainted with “AWS support’s incompetence - or deception,” according to the blog update.

What stuck in Boudih’s craw was the prior insistence of AWS staffers that all his data had been terminated. “The instances were stopped. Not terminated. Stopped.” Stressed the engineer, obviously angry about what he called “gaslighting” by previous AWS contacts. Moreover, his RDS instanced had been backed up as recently as July 19 – several days after support had insisted that everything was “terminated.”

Nevertheless, Boudih refrained from insisting AWS reps were outright lying. There was also the possibility that there was an “undocumented ability to restore ‘terminated’ instances - which would make sense as a safeguard against internal sabotage or mistakes.” There is also the excuse of AWS employee incompetence to consider, but Boudih seems to favor the theory that that prior support contacts “gaslit me about infrastructure.”

According to Reddit threads that spawned in the wake of Boudih’s dizzying initial post, this alarming case was far from unique. The shared payer model has inherent flaws, other social media users appeared to confirm, especially if one party defaults on payments due. This appears to be what caused the software engineer's AWS account and data to be suspended, then terminated.

Account suspension is "part of AWS’s standard security protocols for accounts that fail the required verification," AWS stated in an email to Tom's Hardware. But it didn't explain the apparent subsequent and rapid termination.

In the previous story, Boudih mused whether an account suspension could be escalated to termination, due to some billing issue with a previous shared payment party. That looks quite likely in retrospect.

'Terraforming, not destruction'

Boudih ends his update by summing up the lessons learned from his painful episode, and by offering some advice to AWS.

Firstly, the software engineer is now going to indulge in “Double and triple backups. Distributed across providers. Encrypted with keys I control.” That sounds sensible, and addresses a fair section of the Tom’s Hardware comments section regarding the initial data loss story.

As for AWS, Boudih doesn’t want it destroyed but ‘Terraforming.’ He explains this idea as using simple preventions to mitigate against overreactions to issues affecting customers. Specifically, he suggests that AWS improve its architecture, communication, and support systems to prevent issues that “punish legitimate users.”

Follow Tom's Hardware on Google News to get our up-to-date news, analysis, and reviews in your feeds. Make sure to click the Follow button.