Avleen Vig
@avleen.com
This opens a whole new (to me) world of thought
November 5, 2025 at 10:10 AM
This opens a whole new (to me) world of thought
Can be done. But it requires everyone to understand it's done for the greater good, not to get rich.
And maybe if people didn't think they NEEDED to get rich to live long, comfortable lives, we'd be able to pull this off.
Shit, there I go on the socialism again.
And maybe if people didn't think they NEEDED to get rich to live long, comfortable lives, we'd be able to pull this off.
Shit, there I go on the socialism again.
October 20, 2025 at 9:40 PM
Can be done. But it requires everyone to understand it's done for the greater good, not to get rich.
And maybe if people didn't think they NEEDED to get rich to live long, comfortable lives, we'd be able to pull this off.
Shit, there I go on the socialism again.
And maybe if people didn't think they NEEDED to get rich to live long, comfortable lives, we'd be able to pull this off.
Shit, there I go on the socialism again.
I don't disagree :-)
The issue is that almost every company in existence today is already well down a path where they didn't build this in at the start.
So here we are.
The issue is that almost every company in existence today is already well down a path where they didn't build this in at the start.
So here we are.
October 20, 2025 at 12:25 PM
I don't disagree :-)
The issue is that almost every company in existence today is already well down a path where they didn't build this in at the start.
So here we are.
The issue is that almost every company in existence today is already well down a path where they didn't build this in at the start.
So here we are.
These and other reasons pile up super fast when you start to dig in.
It def can be done. All of these can be worked around. But just saying "it's hard" is a huge understatement. How many years of feature development and bug fixing should a business put on hold to protect against an event like this?
It def can be done. All of these can be worked around. But just saying "it's hard" is a huge understatement. How many years of feature development and bug fixing should a business put on hold to protect against an event like this?
October 20, 2025 at 12:12 PM
These and other reasons pile up super fast when you start to dig in.
It def can be done. All of these can be worked around. But just saying "it's hard" is a huge understatement. How many years of feature development and bug fixing should a business put on hold to protect against an event like this?
It def can be done. All of these can be worked around. But just saying "it's hard" is a huge understatement. How many years of feature development and bug fixing should a business put on hold to protect against an event like this?
Where does your control plane live?
If it's down, can you redeploy it elsewhere? What about the secrets and keys that it needs to manage your fleet? What about deploying code?
If it's down, can you redeploy it elsewhere? What about the secrets and keys that it needs to manage your fleet? What about deploying code?
October 20, 2025 at 12:12 PM
Where does your control plane live?
If it's down, can you redeploy it elsewhere? What about the secrets and keys that it needs to manage your fleet? What about deploying code?
If it's down, can you redeploy it elsewhere? What about the secrets and keys that it needs to manage your fleet? What about deploying code?
You may have legal / compliance / data protection reasons that come into play too. Maybe you can bounce around regions in the same geopolitical location, but what if those aren't available? You may not be able to go just anywhere else.
October 20, 2025 at 12:12 PM
You may have legal / compliance / data protection reasons that come into play too. Maybe you can bounce around regions in the same geopolitical location, but what if those aren't available? You may not be able to go just anywhere else.
Compute isn't 100% fungible. AWS offers many instance types because businesses need to optimize which ones they pick for cost and performance reasons. Those instance types may not be available in other regions in the same quantities.
October 20, 2025 at 12:12 PM
Compute isn't 100% fungible. AWS offers many instance types because businesses need to optimize which ones they pick for cost and performance reasons. Those instance types may not be available in other regions in the same quantities.
If you have large compute jobs that runs for hours or days, they can't be easily interrupted.
If those jobs depend on data stored in one region, it may not be available in another region.
If those jobs depend on data stored in one region, it may not be available in another region.
October 20, 2025 at 12:12 PM
If you have large compute jobs that runs for hours or days, they can't be easily interrupted.
If those jobs depend on data stored in one region, it may not be available in another region.
If those jobs depend on data stored in one region, it may not be available in another region.
If you have customers that pin traffic to specific regions for latency reasons, you can't just bounce them elsewhere.
Or if you have real-time media connections, those may need to persist for an extended period of time or be restarted.
Or if you have real-time media connections, those may need to persist for an extended period of time or be restarted.
October 20, 2025 at 12:12 PM
If you have customers that pin traffic to specific regions for latency reasons, you can't just bounce them elsewhere.
Or if you have real-time media connections, those may need to persist for an extended period of time or be restarted.
Or if you have real-time media connections, those may need to persist for an extended period of time or be restarted.
Almost everyone needs centralised billing for a start. That's OK if you just bill on a monthly subscription model, but if your business takes real-time spend calculations into account ("usage based" models), having that multi-region is a whole different game.
October 20, 2025 at 12:12 PM
Almost everyone needs centralised billing for a start. That's OK if you just bill on a monthly subscription model, but if your business takes real-time spend calculations into account ("usage based" models), having that multi-region is a whole different game.
Georeplication is one thing. Actually being able to run large, distributed, interconnected services entirely independently in multiple regions is something very, very different.
There are really good reasons why so few companies do it well.
There are really good reasons why so few companies do it well.
October 20, 2025 at 11:57 AM
Georeplication is one thing. Actually being able to run large, distributed, interconnected services entirely independently in multiple regions is something very, very different.
There are really good reasons why so few companies do it well.
There are really good reasons why so few companies do it well.
She even said, and I quote, "it could be worse".
WHO DOES THAT
WHO DOES THAT
October 20, 2025 at 11:55 AM
She even said, and I quote, "it could be worse".
WHO DOES THAT
WHO DOES THAT
John is just glad that he doesn't have to any more!
October 8, 2025 at 8:52 PM
John is just glad that he doesn't have to any more!
I feel like your message here could be an infection point for readers
October 6, 2025 at 11:29 PM
I feel like your message here could be an infection point for readers
You too?! I don't feel as bad now!
October 5, 2025 at 8:14 PM
You too?! I don't feel as bad now!
I swear: until very recently I didn't even know they made a movie :-/ we must rectify!
October 5, 2025 at 8:14 PM
I swear: until very recently I didn't even know they made a movie :-/ we must rectify!
Ahhh Google docs. Where status updates go to die
October 4, 2025 at 12:05 AM
Ahhh Google docs. Where status updates go to die
When we realised humans are poor communicators.
Ironic, I know.
Ironic, I know.
October 4, 2025 at 12:01 AM
When we realised humans are poor communicators.
Ironic, I know.
Ironic, I know.