Thread by @swardley

X : You seem dismissive of hybrid cloud.
Me : Quite the opposite, I'm its biggest fan. The use of hybrid public cloud such as combinig us-east-1 and eu-west-1 is an excellent idea.
X : They're both regions of AWS.
Me : They're different regions! Hence hybrid.
X : No ...

X : ... we mean is multiple providers.
Me : Oh, you mean AWS + MSFT rather than different regions of the world? So, so you're adding complexity to protect against failure of either AWS or MSFT ... sounds a bad trade off. Your company is more likely to fail than either of them.

X : ... what about private cloud?
Me : Wow, now you're really adding complexity and cost. To be blunt, in 15 years I've only heard one good argument for private and that was as a transitional play.
X : Who said that?
Me : I did, 11 years ago when it briefly made sense.

A public cloud hybrid of using multiple regions i.e. us-east-1 and eu-west-1 is relatively sensible and adds some complexity but for considerable security benefits I.e. protection against flooding, natural disasters, war etc.

A public cloud hybrid of using multiple global providers i.e. AWS and MSFT is less sensible and adds considerable complexity for the questionable benefit of trying to protect against failure of one of these global giants. It’s more likely the customer will fail than these giants.

A cloud hybrid of public + private makes no sense whatsoever. It adds vast amounts of complexity and cost for no obvious benefit beyond some vague notions of market failure. You would only pursue this path is forced to by legislation.

The arguments about vendor lock-in are mostly codswollop sold by vendors riddled with self interest. How many organisations run around screaming diversify on email to multiple email systems including home grown because of "lock-in" etc. It's just pure garbage.

X : What if AWS switches your service off?
Me : I would always assess the risks. So, if your business plan includes building a neo nazi or terrorist cell or a misinformation campaign backed by a foreign power then in those circumstances a private cloud might be the better path.

X : What about protecting a company's most important asset - data!
Me : "Data is the new oil" is just a catch phrase used to flog huge amounts of storage to customers who then end up with massive data swamps and a CFO asking "When exactly is the value supposed to appear?"

I know companies that produce such vast quantities of data that you could not physically store it all and hence they don't try to. Instead they build adaptive models that learn by being immersed into the flow of data.

On the other end of the scale I know companies that don't realise that data is a form of capital that evolves and what they are often trying to protect became a commodity, often freely available ... long ago.

X : Do you not think that companies will use many cloud vendors.
Me : Of course they will. The cloud vendor I use for cloudy office tools maybe different from my cloudy CRM maybe different from my cloudy storage and compute provider. Of course we will use multi vendors but ...

... for a discrete function i.e. compute then the use of multiple vendors adds additional complexity to which I need to balance the additional benefits.
X : Such as resilience?
Me : I get most of the benefits of resilience by using multiple regions in a single provider.

,,, so I have to weigh the additional benefits (i.e. loss of the entire vendor) vs the costs of training in multiple platforms, translation errors, complexity of managing and monitoring across heterogenous environments, use of lowest common demoninator ... it's a long list.