Private IPv4 allocation becomes much simpler as hardly anything uses it. Ranges can be smaller. EKS doesn't need an overlay network for pod IPs. You can just route everything natively.
Azure's implementation of IPv6 is just IPv4 with extra steps. You still have NAT, and you're charged the same rate for public IPs.
Worst of all, IPv4 is preferred over ULA IPv6, so you're not even using the thing you spent all that effort setting up.
Absolutely backwards.
AWS got this right, automatically assigning a GUA /56 prefix. This simplifies everything. No NAT needed except for IPv4 back-compat. DNS64 is a toggle and NAT64 is automatic. Only public v4 needed is for NAT gateway and ELB.
Best of all, you're actually using IPv6 over v4.
@FraCesc0@bramus So long as you can find a way to override them, you could probably use css variables to define the light and dark variants for the selected theme
@hnasr I think a lot more applications could be run off a single machine if they were designed for performance from day 0. However, perf only becomes a consideration when the app starts feeling sluggish, and it's often too late to get back to optimal efficiency.
@awakecoding@timClicks In a correct system, invalid input would produce a 4xx response, not cause an exception. Granted that assumes a "correct" API and quite often it's far too easy to miss an exception in languages like Java, C# or typescript
@PAYDAYGame People are underestimating the amount of time and effort it takes to diagnose and fix a fundamental bottleneck when your servers receive this much load.
It's also very easy for these bottlenecks to appear out of nowhere. Scaling to immense load is no easy task