Hacker News Reader

Comment by Eridrus | Hacker News Reader

At what odds are you willing to make a bet on this? Do you think there's a 20% chance it's gone in 5 years? Do you think it's 5%? Do you think it's 30%?

People keep saying things will be shut down and I think they should put their money where their mouth is.

View on HN

Even 5% is uncomfortably high if you're planning to build your mission-critical infrastructure around it.

A 5% chance of needing to do a migration in 10 years is very small in comparison to other technical risks.

And lets be clear, this is not a catastrophic risk; your contract with GCP guarantees they will not discontinue services without 12 months of notice so you will have time to do an orderly migration unless Alphabet itself goes under.

I am a GCP user; I have found their offering (actually having GPU quota) significantly better than the other clouds and GCP shutting down does not keep me up at all.

12 months notice is too short for big non-tech enterprises. They plan their infrastructure needs in 5-year terms. Changing to a new environment takes re-educating a lot of people, going through a lot of legal and audit meetings, finding enough manpower to do the transfer, putting things on hold purely for the migration,...

12 months is the time for the contracts to settle between these enterprises and cloud-vendors.

If your company is that big and unwieldy that a tiny chance of needing to do a migration in 12 months is unacceptable, you can stick to the key services which offer 36 months of notice. For all I know at that point you can negotiate your own guarantee if you're so big.

If you work at a company that can't do a cloud infra migration in 36 months, I don't have any good advice for you, but I doubt most commenters on here fall into that bucket. Frankly, I would expect companies that large to be in all the clouds so that you could negotiate them against each other, but I doubt that impacts many people here, and the chance of any of these worst case scenarios happening is vanishingly small.

This isn't a tiny change, if GCloud says "we quit in a year". That is a huge change. Applications using GCloud SDK for services suddenly need to be reworked and redrawn. That alone is a huge change for non-tech companies.

Some companies even need to check with the laws they reside in if the new cloud vendor can even be used.

Just work in any financial, medical, pharmaceutical, government-related... company and you'll understand that a year is nothing for these companies.

What an oddly aggressive comment - perhaps the problem isn't prediction but extrapolation of past experiences with Google, would you not agree that it gives you a sliver of worry that they may cut the service off?

You seem very intense about Google's offering, maybe you feel attacked by the points made in the thread, but the way you're approaching this discussion is very strange.

I'm tired of reading the same thing over and over where people prognosticate based on poor readings of events and I think people should put there money where their mouth is.

I really do not think the chance of GCP discontinuing a core service is high enough to warrant the level of comments this gets and if I find myself constantly reading these comments, maybe I can at least be compensated by people making bad bets or maybe people will admit the chance is really low.

So if you are so worried about this, please give me a number that represents how worried you are about this.

Nonzero is enough to choose an alternative.

Lol, ok man, sounds like you're real good at making rational choices. I'm sure there's exactly zero chance AWS or Azure will ever have any problems.

Do you not consider Google Cloud IOT an important service?

It is not a core cloud service, it is a service for a niche industry that never met the hype it had and probably didn't get many customers.

I don't think discontinuing cloud iot provides any indication that compute engine is going away.

AFAICT cloud iot core was hosted mqtt, which is a thing others offer and you can transition relatively easily. I don't think the shutdown of it was catastrophic for any companies; the quotes I saw in articles were of the sort "yeah, that's annoying, but it shouldn't be hard to migrate to something else".

Eridrus used to work at Google. Not sure if it would be good to disclose before aggressively asking people to bet on the lifespan of Google products, I certainly would err on the side of disclosing.

Sure, I used to work at Google, I used to work at Microsoft, some of my best friends worked at Amazon, I own stock in all of them.

Are you interested in making a bet on the chance GCP will be discontinued?

Reply on news.ycombinator.com