,
20 tweets,
4 min read
Read on Twitter
The cutting edge of "cloud" stopped being about networking layers about 4 years ago
Back when "cloud" was first a thing (you know, before 2010) what we had was instances, and these were basically a virtualised "server", so networking was required.
That meant we could transfer all our knowledge of real world servers from Data Centres to...
You guessed it! The "cloud"
You guessed it! The "cloud"
What all this networking and instances meant was that we could do that wonderful thing...
The "lift and shift"
Because if you've got a data centre, with a bunch of computers, you could easily move it to the "cloud" because it's just a virtual equivalent.
Simple!
The "lift and shift"
Because if you've got a data centre, with a bunch of computers, you could easily move it to the "cloud" because it's just a virtual equivalent.
Simple!
So...
Lift and shift
Lift and shift
Lift and shift
"Cloud" is awesome isn't it?
Well...
Lift and shift
Lift and shift
Lift and shift
"Cloud" is awesome isn't it?
Well...
The thing is that cloud vendors, the ones who were actually innovating in this space, actually looked at networking and thought...
"Why do customers need to do all the networking?"
"Why do customers need to do all the networking?"
Good point. Networking is actually really fiddly. Also, if you are running a service, like data storage, you can simply do the networking for them. Just provide them with an internal address, and voila, the networking is done for you!
And if you can abstract your networking with things like identity management and access management in various forms, then... you don't need to worry about networks.
So then talking about "cloud" in terms of "networks" becomes redundant.
So then talking about "cloud" in terms of "networks" becomes redundant.
Because "cloud" is then about services, and security, and not about networks.
Not to say that networks are irrelevant. If you're using something that needs a network, like a connection based RDBMS, then you can still do all the things like before, but it's not required.
Not to say that networks are irrelevant. If you're using something that needs a network, like a connection based RDBMS, then you can still do all the things like before, but it's not required.
And that's where the cloud vendors have been going *for years*.
The weird thing is that many technologists still think that "networking" is a requirement for doing "cloud".
It isn't.
It hasn't been for years.
The weird thing is that many technologists still think that "networking" is a requirement for doing "cloud".
It isn't.
It hasn't been for years.
If you think that doing "cloud" means that you have to understand VPCs and networking and things like that, then you're probably going to teach people the skills of yesterday, and not the skills of tomorrow.
The skills of tomorrow are identity and access management, security, understanding how to manage configuration as code (terraform, cloudformation etc) and making solutions that are easily composable and sharable.
Oh and yes, this is what serverless is all about.
It's why I call it "cloud 2.0"
I did a talk on it in April in Atlanta
recallact.com/presentation/s…
It's why I call it "cloud 2.0"
I did a talk on it in April in Atlanta
recallact.com/presentation/s…
So, if someone is telling you that "cloud" is just someone else's data centre, then, to be honest, that person is stuck in a different decade.
Also, there's no easy way to do multicloud in the scenario above, because while you can abstract networking and "instances" relatively easily, and it's relatively simple to think about kubernetes style solutions in this way, it's definitely not simple to abstract config and IAM
The issue with networking being intrinsically linked to "cloud" is why I think kubernetes is so popular with people. I think it's also why people tend towards kubernetes rather than configuration and IAM, because it feels more like you're in control.
There's nothing really wrong with that, except that that's not necessarily the purpose of a techie. The purpose is about business value. It's not about how comfortable you are with a technology. Sometimes you need to learn new things.
So when someone says that "cloud adoption models continue to point to multicloud or hybrid cloud" then to be honest, they are simply spouting the marketing.
I've seen almost zero successful multicloud strategies (no... actually zero)
I've seen almost zero successful multicloud strategies (no... actually zero)
Hybrid cloud? I have more time for that, but only in larger companies, with multiple teams, where you pick the right tech for the right reasons.
Only regulatory reasons should require a multicloud strategy, and even then it should be a last resort.
Only regulatory reasons should require a multicloud strategy, and even then it should be a last resort.
I know this is a rant, but I keep seeing people talking about "cloud" when what they really mean is "IaaS" or really "virtual data centre".
If you want to talk "cloud 2.0" then come talk to me.
If you want to talk "cloud 2.0" then come talk to me.
