Hi, I’m a systems and network admin, who loves to speak of and peek into infrastructure designs, and I was wondering if I could get any info on Mondo’s infrastructure for handling card payments, and the reliability thereof.
According to news sources the incident with people not being able to use their cards on that bank was linked to the outage on AWS services.
So I wonder, from the banking experts here at Mondo, how does it work, when an ATM, card machine or whatnot queries for a transaction approval, how does that happen? (Protocol-wise)
And, can an outage like this happen with Mondo as well, provided an underlying service outage in a single, region happens? (If Mondo uses AWS in the first place. I thought I read somewhere they do.)
Hey Marco, Our infrastructure is something we’re really proud of. It’s something we decided, from day 0 to get right, and we’re still working on it. (and it’ll never be done done)
Under the covers, we’re using AWS, with Multi AZ everything, and our plans include having warm databases syncing to multi-regions as well.
As you’ve probably seen in Matts talk, we’re building a on a microservices platform, last time i checked our system was over 100 services, all written in Go.
To schedule these we’ve been using Mesos and Marathon, which was a good choice a year ago when Docker was less mature, but we’re currently looking at/working on moving to Kubernetes and containerising everything.
Our back end engineering team likes to joke we’re building a distributed monitoring and tracing system that happens to do banking.
That’s good to hear that infra is not being left out, I like what I’m seeing, and look forward to possible blog posts and interesting tech blogs, if that’s something you plan to do.
Multi-AZ is a must in HA-AWS services, it’s good to hear you make use of them!