They're pretty readable, I recommend that second one (though it leaves some questions I have unanswered). It's a good example of how latent issues with a large system can go unrecognised for a long time before a seeming unrelated change can trigger a cascading failure