dward

Data protection policies may apply to the hard drive.

It's to be seen but owners of the spec represent mesos, kubernetes, docker and cloud foundry so it seems like Docker is open to the idea. source:

https://github.com/container-storage-interface/spec/blob/mas...

If anyone were to have those, it would be google.

Last time I tried to use keybase it required me to paste my private key into a browser before using any of the advanced features like chat. This seems unnecessary and doesn't make any sense from a security perspective. Has the situation changed? I won't use their service until this is fixed.

The producer closes the channel. This is differentiable from an open empty channel in a select.

In my opinion, yes.

But even those aren't allowed to be mounted to your corp laptop. Those clients can only be mounted on a workstation unless you are doing iOS development.

Yes chromebooks are very common developer laptops, up there in numbers with (still most popular) macbooks and linux laptops. They are more than sufficient for the lion's share of engineering roles. Corp engineering is pushing them hard because they are cheap and secure.

This is not common practice. Google has had a strict no iCloud, no Dropbox policy for a while. We can't even have code from the core codebase stored locally on our company issued laptops.

It is far harder to exploit on Kubernetes then on a laptop running a dnsmasq cache attached to a starbucks wifi.

You would need to middle man traffic between dnsmasq and upstream DNS or have some control over DNS records and be able to force dnsmasq to do lookups. As you point out the escalation is also far more limited. Ok, so you have a shell in the dnsmasq container. Now what?

Number two is to throw in your safe and forget about.

Google Cloud SQL supports postgres as well.

With Alibaba cloud, infrastructure like databases is their business. Take AWS Aurora and GCP Spanner as examples.

One of the original goals of Google's QUIC[0] protocol was to achieve 0-RTT handshake. I wonder how TLS 1.{2,3} Session Tickets will affect the future of that protocol. Has it achieved it's purpose (applying the right pressure to the TLS working group)? Will it be phased out now?

[0] https://blog.chromium.org/2013/06/experimenting-with-quic.ht...

Crouton is just a chroot to some linux distro. It doesn't use any sort of virtualization.

What happens if the process can't shut down cleanly and writes are in flight? Is there a way to check for hash table corruption? Is there a plan to support deletion/compaction/defrag? Why would I use this over something like LevelDB?

CMU doesn't require, but it does allow:

https://www.sei.cmu.edu/downloads/sei-cert-c-coding-standard...

No it hasn't. You can track the progress of CronJob by subscribing to the feature issue (which should generate fairly low email traffic):

https://github.com/kubernetes/features/issues/19

Out of curiosity, are you also using the cloud CDN?

https://cloud.google.com/compute/docs/load-balancing/http/us...

Ripped it out implies that at one point they used it.

Curious, why do you think they have more servers than google? A quick search didn't turn anything up.

I'm asking for specific examples.

What services are missing from GCP?

> If by google's age and experience, you mean the requirements of the Google Cloud, then i agree.

Why do you think this is a requirement of Google Cloud?

From the tcmalloc paper[0]:

> Speed is important for a malloc implementation because if malloc is not fast enough, application writers are inclined to write their own custom free lists on top of malloc. This can lead to extra complexity, and more memory usage unless the application writer is very careful to appropriately size the free lists and scavenge idle objects out of the free list.

An example of such an optimization is the arena allocator[1][2] employed by protobuf. Custom memory management schemes are not uncommon in performance critical code.

0: http://goog-perftools.sourceforge.net/doc/tcmalloc.html

1: https://en.m.wikipedia.org/wiki/Region-based_memory_manageme...

2: https://developers.google.com/protocol-buffers/docs/referenc...

Yes. The only way to delete things at scale is to ship them over the network to a dedicated cluster of machines running a proprietary fork of /dev/null. Google has invested man decades into this system and holds numerous patents on the design.

Yes, it runs both systemd and upstart. Staying true to chromiumos, upstart is still the /sbin/init. I believe both upstart and systemd support some subset of the sysvinit api (e.g. rc and /etc/init.d).

The previous debian based container-vm image only ran upstart.

Probably should link this as well: https://github.com/coreos/locksmith

Don't they do this every release?

Google as well?