hanikesn

The correct url is: https://aws.amazon.com/blogs/compute/improving-performance-o...

Symfony really paved the way for quality PHP code. Excellent Web framework which is a lot of fun.

> My only real gripe with PHP is the annotation syntax. Yikes. Stuff that's in comments SHOULD NOT AFFECT RUNNING CODE. Who came up with that?

It's a workaround to copy Java's annotations which IIRC also came from javadoc. PHP-8 is adding proper annotations as well. In general being able to add parseable meta information to your code can be quite useful.

I switch PHP to C to show OP in a mirror how ridiculous his statement is.

Once or twice a year I have to peak at the code of a C app that was developed by interns more than 10 years ago. No framework was used, max 3 letters variables are the norm, no blank lines and the only comments are from code copy pasted from examples.

These days I would be more inclined to celebrate the death of C than its birthday !

I think I/O heavy benchmarks would be much more useful as syscalls take the heaviest hit from the mitigations.

His application runs for a total of 3 ms. So startup time is not the issue.

The exact same thing happens hard drives. There's a ton of magic going on.

Yes, it does. It's still kube-proxy though, but it will use ipvs instead of iptables.

I don't think I ever heard of anyone (besides scientists) to store timestamps in TAI. Though Ada supports leap second calculations in the standard library: http://www.adaic.org/resources/add_content/standards/05rat/h...

Most applications can't be easily condensed done to a single static binary.

(Un)fortunately that's done by a third-party provider. Fun-fact: Bad wifi and delays have the highest impact on customer happiness.

Disclosure: I work for Flixbus

It's a bit more complicated. We have a revenue share model with our partners, payout depends on a few factors and is about a 70/30 split. Though we also guarantee most of them a minimum payout per ride. So it's in both parties interests to have satisfied customers.

The bus terminals are owned by the cities themselves and only at the busiest stops we have those booths. Also a lot of our manpower goes into planning new lines, marketing and customer service. And not having to buy buses at 275k € each allowed us to scale up business really quickly.

Also we really see ourselves as a technology company and are always looking for more engineers. So if you want to work in Berlin or Munich in an international team drop an email to jobs (at) flixbus.com.

Disclaimer: I work for Flixbus as an SRE in our cloud platform team.

They also announced a new $400 permanent price for the Rift+Touch.

This is not an acquihire. A lot of developers were already let go quite some time ago from Basho.

I guess no one here reads https://blog.fefe.de/

It's the only page you can read on a GPRS connection (http) and you can literally see every packet as it's transmitted, because the page is rendered bit by bit.

We're using Puppet to deploy our kubernetes cluster on bare-metal as well as on AWS.

Achieving HA isn't necessarily complicated. Etcd supports clustering by default and the master components have master election builtin.

We also use haproxy locally on every machine to loadbalance between the different api servers. So we don't need a central LB.

Setting up HA was only a small part of our overall effort. Making things robust and figuring out the small details are a lot harder.

What about using kubernetes for running gitlab ci?

How does the configuration look like for a bare metal setup? E.g. spreading pods over multiple racks?