global-microsoft-outage

blackbird9 said:
Petrol generator, lithium house battery, power packs, 3 FM radios to pick up emergency broadcasts, spare smartphone, big packs of duracells, water for 2 weeks, lots of cans of beans and soup, emergency chocolate, calor gas heater, camping gas stove, first aid kit, more batteries, crappy wind-up radio/torch thing, sub-zero rated sleeping bag, more emergency chocolate and biscuits, extra jumpers, coats, gloves, hats and duvets, 500 bog rolls just in case, emergency flashlights, old car battery in the garage hooked up to old car radio, etc, etc... I guess it must be winter approaching. Its all good, like caffeinated soap.

View attachment 27905View attachment 27907View attachment 27908

I guess the toilet-roll buying here is not quite at the full-on pandemic panic level yet (or 'bog-roll-gate', as we call it) 😬

View attachment 27909
Click to expand...
Some Bosnian had nice blog:
Besides clean water and antibiotics, he mentioned disinfectants (any little cut can get septic fast) and the most interesting thing:
stash of false gold "wedding" rings to use at road blocks as bribe.

And the most important thing to have, so as not to become supermarket, especially in multi-ethnic societies:
a gun and a lot of ammo.
 
hedwards said:
Absolutely, but not being tied to a cloud company is still something to avoid.

I've known people that work at MS and a real problem is that they literally can't fix bugs because they don't have a test lab and the software is so bloated that often they can't even get a full core dump to even start identifying the instructions that caused the panic. So bugs that occur at MS can and do get fixed, the others are often times unfixable as they don't know what happened.
Click to expand...
Their most popular product Active Directory is also buggy as hell when you have many sites replicating over faulty or slow lines.
Some guru from Microsoft had to fly over and tailor it to our environment.

Monolithic kernels are probably/hopefully the way of the past - at least on bare metal. Hypervisors gonna be microkernels with drivers in userspace i.e. no kernel panic when they fail.
Monolithic OS can then run on predictable virtualized HW for speed and legacy reasons.
GUI advantage is most likely obsolete too - even now it is pain to automate and with agentic Ai, you don't need GUI, just proper API.
 
DanielBowman said:
Some Bosnian had nice blog:
Besides clean water and antibiotics, he mentioned disinfectants (any little cut can get septic fast) and the most interesting thing:
stash of false gold "wedding" rings to use at road blocks as bribe.

And the most important thing to have, so as not to become supermarket, especially in multi-ethnic societies:
a gun and a lot of ammo.
Click to expand...

Ammunition is the thing to hoard. It will be excellent trade material by itself.

Just beware secondary explosions when your place is hit.
 
cracauer@ said:
Ammunition is the thing to hoard. It will be excellent trade material by itself.

Just beware secondary explosions when your place is hit.
Click to expand...
Yeah, it was my best investment (on paper - not gonna sell).
Not sure about the secondary explosions though.
It is just loud bang, some shrapnel but not much penetration force. The weapons chamber is what makes the bullet fly fast by directing the blast, enclosing the casing.
 
  • Like
Reactions: mer
DanielBowman said:
It is just loud bang, some shrapnel but not much penetration force. The weapons chamber is what makes the bullet fly fast by directing the blast, enclosing the casing.
Click to expand...
The US Military did a lot of tests around this. Also data in "Hatcher's Notebook". A 22LR laid on a rock and smashed with a hammer acts as if it's chambered, so maybe it's more "containment" than "chambered"
 
T-Aoki said:
Thankfully, I don't have Micro$oft account (except quite old registration for DOS and Windoze 3.x). So I'm NOT affected here.

But seeing this (and outages of other clouds), I strongly believe ALL clouds SHALL have 100% same API (internal implementations can be 100% different, though).
If this is MANDATED by international law, anyone (individuals and / or organizations) can sanely have backup environment that is always except outages in sync and switch to another if one fails (with additional costs, of course) if they consider the system important. For mission critical systems, more than 2 cloud providers can be used and kept in sync.

Current APIs should be supported in conjunction with mandated compatible API to allow users "slow migrations" at least for several years.
Click to expand...
Not a good idea, because it is not necessary. Today you can download libraries that make common cloud services go through a common API. Since I'm more interested in storage, I've used such a library to test out storing interchangeably in S3, GCS and Azure. To get what you want (interchangeable backup on multiple providers) is already easy.

Bad idea, because it stifles innovation. If a company (including a cloud provider) has a good idea for improvement, it should not be hamstrung by having to use a legislated API.
 
DanielBowman said:
Yeah, it was my best investment (on paper - not gonna sell).
Not sure about the secondary explosions though.
It is just loud bang, some shrapnel but not much penetration force. The weapons chamber is what makes the bullet fly fast by directing the blast, enclosing the casing.
Click to expand...

Well, not if you hoard 81mm HE mortar rounds...
 
DanielBowman said:
Some Bosnian had nice blog:
Besides clean water and antibiotics, he mentioned disinfectants (any little cut can get septic fast) and the most interesting thing:
stash of false gold "wedding" rings to use at road blocks as bribe.

And the most important thing to have, so as not to become supermarket, especially in multi-ethnic societies:
a gun and a lot of ammo.
Click to expand...
Yeah, we have it easy over here. So far, anyway. Let's hope it stays that way.
 
cracauer@ said:
Well, not if you hoard 81mm HE mortar rounds...
Click to expand...
I haven't got any ammo, mortar rounds, etc. I have got a chain saw though. It's a nice Stihl petrol one. And I've got a Honda rotary tiller, you could do some damage with that, those tynes are vicious.😬
Really great little tiller, 35 years old and still going strong, leaks a little bit of oil but otherwise fine. Just don't get your foot near the tynes...
1794677667452.png
///
 
ralphbsz said:
Do you have any data to show that your self-hosted servers are more reliably than cloud services?
Click to expand...
Did I ever use the word "reliable" in my OP? I just said I've been raging against cloud infrastructure for years, and didn't state my reasons, but like a certain important historical document "we holds these truths to be self-evident". LOL

And in quantifying the argument in an online forum...I think we both know it's a lot more complicated than simple up/down time stats. There are other issues at play that are often less quantifiable: such as data security, confidence level, loss of control over private data, changing market whims in network companies, 3rd party meta-data collection and tracking...To me there are many good reasons to consider self-hosting services. "Reliability" is just too broad a term use when other considerations should also be valid.
 
ralphbsz said:
Do you have any data to show that your self-hosted servers are more reliably than cloud services?
Click to expand...
One or two self-hosted servers are not more reliable, but from 10,000 self-hosted servers most time will work more than 90-95%. Cloud service is more realiable than most of self-hosted servers separately but not from their whole "group" (if we see them as group). The only way to stop more than 40% of them is some really critical bug existing in any OS or disconnection of network.
 
blackbird9 said:
Petrol generator, lithium house battery, power packs, 3 FM radios to pick up emergency broadcasts, spare smartphone, big packs of duracells, water for 2 weeks, lots of cans of beans and soup, emergency chocolate, calor gas heater, camping gas stove, first aid kit, more batteries, crappy wind-up radio/torch thing, sub-zero rated sleeping bag, more emergency chocolate and biscuits, extra jumpers, coats, gloves, hats and duvets, 500 bog rolls just in case, emergency flashlights, old car battery in the garage hooked up to old car radio, etc, etc... I guess it must be winter approaching. Its all good, like caffeinated soap.

View attachment 27905View attachment 27907View attachment 27908

I guess the toilet-roll buying here is not quite at the full-on pandemic panic level yet (or 'bog-roll-gate', as we call it) 😬

View attachment 27909
Click to expand...
That made me chuckle. I remember the bog roll frenzy days. My biggest chuckles came from seeing the solo car driver occupant all masked up to the nines. It took a long time for the panic to die.
Reminds me - add watch vids on 'mass formation psychosis' to TODO list hehehe.
 
6502 said:
One or two self-hosted servers are not more reliable, but from 10,000 self-hosted servers most time will work more than 90-95%. Cloud service is more realiable than most of self-hosted servers separately but not from their whole "group" (if we see them as group). The only way to stop more than 40% of them is some really critical bug existing in any OS or disconnection of network.
Click to expand...
I think sometimes it's "if it fails who owns it".
Self hosted may not be more "reliable" but removes external dependency. Yes there is a cost of needing the specialized knowledge inhouse, but often easier for "corporate" to lean on internal resources vs external. We've all dealt with 3rd party customer service; how much leverage do we actually have on them? Call "drops", breaks up, my word vs their word.

In the past I've had customers complaining about our product not working but then find out that someone didn't call DigSafe and broke fiber cables. Way outside our sphere of responsibility, but customer first escalation was to us. Took a while to get all data and have all the meetings before "Ok, not your problem but we're still keeping an eye on you".

This specific MS outage (based on current info), if a company self hosted DNS, Mail servers, used locally installed software instead of "cloud based sw", I think there would have been minimal impact. People could still work on documents, people could still send email to non-ms endpoints.

But part of the cost would be "local company needs to have expertise to deal with these things". Most of the time, it would run by itself, so look like a cost sink, but in the corner case, would save the company.
 
ralphbsz said:
Not a good idea, because it is not necessary. Today you can download libraries that make common cloud services go through a common API. Since I'm more interested in storage, I've used such a library to test out storing interchangeably in S3, GCS and Azure. To get what you want (interchangeable backup on multiple providers) is already easy.

Bad idea, because it stifles innovation. If a company (including a cloud provider) has a good idea for improvement, it should not be hamstrung by having to use a legislated API.
Click to expand...
NOT at all.

Once becoming "infrastructure", stabilitiy, compatibility, "non-vendor-lockin" should be far, far and far over innovation. And as you can see within this panics, clouds are already "infrastructures".

A corporate can vanish, without inheriting their services to any successoros.
And data are too important to be allowed to vanish with a vendor.
(Backward) Compatibilities are MANDATORY over innovations!

The exactly same can be said for programming languages. It's infrastructures, too. See the recent panics with PHP.
100% backward compatibilities are MANDATORY. SHALL not be avoided.
 
At the end of the day; no one has an answer Active Directory. That’s the elephant in the room here, and why most organizations still use Windows.
 
ralphbsz said:
Do you have any data to show that your self-hosted servers are more reliably than cloud services?
Click to expand...
Depends what you mean, with self-hosting you've always got the data available more or less no matter what. You just have to make sure that it's backed up. If you mean not breaking randomly, I haven't really had much trouble with that either, and I can schedule downtime for whenever it's convenient to me, not the owner of the hardware.

I generally have the plan b of SSH so long as I have a connection at all.

But, there is stuff like the internet connection and not having multiple mirrored servers that you often get with the cloud, along with a bunch of vendor lock in.
 
You must log in or register to reply here.
Back
Top