It should be around 25 ms in normal conditions. That's what I got when pinging Hetzner in Germany, from Finland, when the cable was still in use and when using a connection that routes through the cable.
I am don't use Hetzner, but I use ssh between Finland and Germany every day. As a matter of fact even back and forth because of tunneling. After reading the news this morning (Hetzner incident is date 3:30 UTC) I was surpised that I had not noted any lag. It remained very reponsive all day.
I have a persistent VPN tunnel between Finland and Germany and I’ve not noticed really any disruptions. If it had cut out for even a moment, it would’ve interrupted my services (since they don’t recover gracefully at the moment) and I would’ve found out.
Specifically this is being pushed by Ylva Johansson [1] from Sweden, who has (reportedly) financial connections to the organisation Thorn which is hoping to sell this chat monitoring software.
> Only the Commission can propose new legislation.
That is a technicality relying on a shallow look at the word "propose". The commission frequently takes direction from the council when deciding what to focus on which leads to them "proposing" legislation. In this case the push for this has come from the Council and certain national governments combined with a particular commissioner.
The Commission agenda and mandate is set by the Council that nominates it, and periodically reviewed by the same Council. Items are set in meetings, the agenda of such meetings is typically public.
If the Commission pushes, it's because the Council told it to push.
The Commission is formed by commissioners, each of which has been nominated by a different nation's government. The Council consists of actual members of the national governments.
The third party that can't propose legislation but has to approve it (and which strongly opposed this) is the Parliament, which is directly elected.
OK, this is probably some nuance of the English language that I'm missing as a non-native speaker, but I meant that the people that make up the Commission are not part of the national governments. The people that are part of national governments each get to nominate one commissioner though, in addition to being part of the Council.
m5 instances use the Nitro system. In addition, m5.24xlarge is a quite quirky instance type: It uses 2 CPU's with 24 cores each in a NUMA configuration. Half of the RAM is attached to each CPU, and access from the other CPU is much slower. In addition, the CPU cores use a microarchitecture from 8 years ago, meaning the cores are quite slow in practice.
All of this means that a lot can go wrong when running code on those instances, resulting in lower performance. It is either advised to run separate processes on each NUMA domain, or use NUMA aware code (which Java almost never is). In addition, the code (or the system) should be highly scalable to multiple CPU cores.
In addition, the cores are old enough to suffer from Spectre/Meltdown related patches/workarounds, decreasing especially syscall performance.
In our case the instance type is about the only workhorse for the given job. High TPS (scales well to the core count) and needs a large on-disk configuration for low latency key value retrieval of data deployed on disk.
I did slightly misspeak on the instance move having seen your reply. We moved from m5.24xlarge to m6i.16xlarge. Sorry for the confusion.
That said, you shared some interesting information. I'd love to read up more on this, any specific place I can dig in a bit deeper regarding the finer specifics of these instance types and architecture?
Yes, I'm aware. The service in question wasn't easily able to be moved so we moved to m6i which isn't ARM based but does leverage nitro. We saw substantial improvements in that configuration too. Not sure what is different because you said m5 use nitro as well but my assumption was m6i with reduced hypervisor overhead from nitro was why we saw improvement.
m6i is a much newer CPU architecture, based on Intel Ice Lake rather than Skylake. It is quite significantly faster just from that alone. In addition, the CPU has about 10% higher clock speed.
The 16xlarge version is also a 32 core single socket CPU, meaning there should be no issues with NUMA. I would expect it to be much better than m5.24xlarge in most applications when taking the much faster single-threaded performance into account. Of course, nothing beats benchmarking and measuring yourself though.
I have personally seen issues with NUMA systems and code that theoretically parallelizes very well. Any synchronized mutable state becomes an issue with these kinds of systems. For example, I have had an issue where third party code would use the C "rand" function for randomness. Even though this was not used in a hot code path, on m5.24xlarge >90% of the execution time would be spent on just on the lock guarding the internal random state. On a "normal" system with fewer cores this never showed up while profiling.
Modern cars are much better with moose crashes. A couple of years ago I was in driving back from grocery store, just south of Sodankylä in northern Finland, during the darkest times of winter, when the car in front of me crashed into a moose. The moose bounced multiple meters into the air from the windshield of the car in front of me, and I needed to do the typical evasive maneuver (https://en.wikipedia.org/wiki/Moose_test) to avoid the flying moose. I'd estimate that the crash happened with speed of around 40-50 km/h.
The person driving the car in front of me survived unscathed, without any injury. The car (2010's VW Passat) very likely was a loss though. This just tells how good modern cars are with these kinds of crashes.
In Finland there are about 1000-2000 moose crashes every year, but only around 1-4 people die from them.
