Slashdot Mirror
Red Hat and Microsoft Partner On Azure (redhat.com)
An anonymous reader writes: Satya Nadella has made some interesting reforms to Microsoft. Today, Red Hat and Microsoft announced that they will partner to deliver Red Hat's product suite in Azure. Red Hat will also support .NET core in RHEL. Additionally, Red Hat's CloudForms product will now work with Hyper-V/Azure, RHEV, VMware, and AWS. Microsoft has certainly come a long way from the Halloween Memos. Here are Red Hat's blog post and Microsoft's blog post about the announcement
A) There has been hypervisor cooperation on both sides for ages so that each other's products ran on each other. This makes sense since 1) it avoids accusations of monopoly behavior 2) it create's a duopoly locking out other smaller competitors.
B) Remember Microsoft not so long ago ended up contributing to Linux.
C) If you want to Embrace, Extend Exterminate then you first have to Embrace. The article is wrong. This is a move straight out of the Halloween.
Hell will freeze over when Microsoft starts actively and fairly contributing new technologies to proper copyleft projects. Not when it works to maintain the traditional IT corporate stranglehold.
(1) do everything poorly, but ship soon, and if you can't ship, bullshit and ship later
(2) get contracts with everyone, do whatever it takes but get the contracts, let some pirate your stuff if they want to just as long as they're running it so maybe they can buy something later
(3) stay in business. profit where you can, funnel the cash into places where you can't profit yet
whereas Apple is
(1) sell luxury products
(2) people who buy Apples do it to supplement their image, and defend the Apple brand in all forums in order to make themselves look good
(3) profit. lots and lots of profit.
and Linux is
(1) make cool stuff
(2) people use it and extend it. Developer users, professional and hobbyist, rave about how nice it is, because they feel a sense of shared ownership of the code
(3) The best extensions are added to the base, in small pieces that each do one thing really well
Cue the comments about angry people switching from RedHat to another Linux distro.
I switched years ago. I'm not angry, but redhat just fell behind in being good for what I wanted.
From what I remember, of digging through the init scripts, it's not surprising that systemd came out of Redhat. A good part of it is meant to speed up booting. Certainly back then, the people at RedHat coldn't write shell scripts for crap. The boot scripts were terrible convoluted messes. No wonder it booted slowly.
I actually cleaned up the X11 start script hugely, because one of the features I wanted was actually completely unreachable after they'd essentially rewritten it 3 times from 5 to 5.2 to 6, and then concatenated all 3 versions. I submitted a bug report and patch which went into a black hole.
I don't see any pressing reason to switch back to redhat any time soon.
SJW n. One who posts facts.
We're in the middle of the planning for the Windows 7 to 10 transition, and 2008 R2 to 2016, so we're getting plenty of face time with the premier support guys. The message is abundantly clear -- Microsoft is done selling one-off licensed software. Everything is going to be Azure based in their mind, and on-premises installations of software are the exception now. Server 2016 has so many Azure hooks that it might as well not have been released as a standalone product. Windows 10's updating model relegates stable releases to a much more minority position than they were in the past...it requires an Enterprise Agreement/Software Assurance to deploy Windows 10 LTSB and avoid constant cumulative upgrades.
In an environment like this, where they're moving back to mainframe style custodial IT service models, why wouldn't they partner with Red Hat or any other OS vendor for that matter? They want companies to move everything into Azure, not leave some bits hanging out on-premises or with another cloud provider. The Windows vs. Linux wars are cooling off because vendors sense the juicy returns in the cloud. Why sell software once when you can force businesses to pay over and over again for decades to use your resources/products? I've said before that both Amazon and Microsoft are building their clouds on the backs of Bubble 2.0, so funding is plentiful and therefore prices are incredibly cheap. The thing to watch will be when the bubble bursts, and a duopoly exists...will those low prices continue?
OpenBox, whatever it actually is, wasn't forced into Debian, Ubuntu, and pretty much every other major Linux distro completely against the will of the users of these distros. That's a big part of the reason why people don't dislike OpenBox. People don't get angry with something if it doesn't cause them any problems. But people do get royally pissed off when something totally unwanted is basically forced up their rectums. That's exactly what systemd was like to many Debian users: repeated, forced penetration of one's anus with an object that's large, sharp and rusty. It hurts a lot to have Debian systems that were first installed years ago, upgraded flawlessly many times over, only to have them ruined unnecessarily when systemd was forced upon Debian users. We told the Debian maintainers that we didn't want systemd. We saw the problems it had caused for so many others, and we didn't want to fall victim to it, either. We told them again and again and again that we wanted no part of systemd. We didn't want it on our systems. We didn't want it even considered for inclusion into Debian. Yet we were treated like dogshit. We were treated like scum. We had systemd forced onto our unwilling Debian installations. It's really disappointing when a distro like Debian, which was pretty much flawless for years on end, even when using its unstable version, suddenly becomes less reliable than Windows ME all thanks to systemd. It's like your internal organs are being torn out when your Debian installation may start to randomly fail to boot after doing what should be routine updates. It's a pain that nobody should ever feel.
I have something like the following inittab fragment that I built on my production servers:
ds:4:respawn:/home/prog/schedule.sh
da:4:respawn:/home/prog/alert.sh
cx:4:respawn:/home/prog/update.sh
cx:4:respawn:/home/prog/audit.sh...
These shell scripts mostly set a number of environment variables, then exec a runas.c program that I wrote that knocks the privilege down from root. After privilege is dropped, my runas program calls exec() on the *real* program that I want init to respawn.
This works, but it's a big pile of duct tape and bailing wire. I'm not proud of it.
I can get rid of all of that stuff with systemd, and launch it correctly:
$ cat /etc/systemd/system/broker.service
[Unit]
Description=broker
#After=network-online.target
Wants=network-online.target
[Service]
#ExecStartPre=
#ExecStopPost=
Environment=ORACLE_SID=mydb ORACLE_HOME=/home/oracle
ExecStart=/opt/pkg/broker
WorkingDirectory=/tmp
Type=simple
KillMode=process
Restart=always
User=nobody
#Group=nobody
[Install]
WantedBy=multi-user.target
I realize that people don't like the dbus integration, the replacement of su with systemctl shell, and many other complaints. However, this code has been carefully designed, it's reliable, and it gives me the ability to throw away a lot of my ugly glue. Call me heretic, but yes, I like it.