Agreed. You shouldn't have to explain someone's problem to them. Ideally when you present a good solution they will immediately match it to a pain point they're already experiencing.
back
1 comments
To be fair, we didn't try to go around selling businesses on "decentralization idealism"; that would obviously be silly.
Plenty of enterprises already need to self-host services, for compliance reasons (ITAR, FISMA, HIPAA, FINRA, GDPR, national data locality laws in Germany, Russia, China, South Korea, and others), or sometimes even just paranoia. The need is very much there.
But, like, we literally didn't know where to start. Pick up a phone and just, like, call people? Apparently that's how sales works but I'm sure as hell not the person to do it. I hate it when people call me! How could I call someone? So we didn't call anyone, and mostly hoped that fans on Hacker News would go sell Sandstorm to their IT departments for us. That was dumb and didn't work.
There's a bottoms-up path that some enterprise saas companies have taken (Slack, Airtable come to mind) where employees/departments just start using it to get stuff done and eventually IT finds out about it but by that point it's ingrained enough where IT signs a contract just to try to rein in all of the one-off spending and data leakage. But for that to work I think the apps would have to have above-and-beyond appeal to individual employees whereas I think a lot of sandstorm's appeal is to the organization itself.
Yeah, exactly. We tried to follow that model, but Sandstorm really wasn't prepared for it, because Sandstorm apps weren't actually better than cloud-based alternatives and so there wasn't a motivation for employees to adopt them.
> To be fair, we didn't try to go around selling businesses on "decentralization idealism"; that would obviously be silly.
Sorry, my comment should have been more clear. I wasn't referring to Sandstorm or you in my comment and I didn't mean to accuse of you of such silliness. It was more a reaction to the poster above who said they were targeting "small business" without specifying which kind of small business and what they are solving for them.
Anyways, yes, there are many reasons for businesses (even small ones) to self-host. Other reasons include the ability to customize (you can't customize a saas, unless you're a gigantic customer), and longevity (Saas come and go, and switching costs can be high).
Yeah I definitely wasn't referring to sandstorm either.
Businesses may not specifically want "decentralization idealism", that's true. What they want is "Windows for Web".
Something where you could install an app and bingo, it instantly worked on all endpoints around the globe.
Or where you once install "yet another CDN provider" component (app), and it instantly and automatically covers all the installed apps.
Kubernetes came close to this, but it requires significant amount of expertise, and honestly it is a matter of time until it gets smashed with something better.
In short, what we (businesses) want is "Click and serve". Sandstorm was so close and yet did not make the expected steps in that direction. This makes me sad, but this is life.
>Kubernetes came close to this, but it requires significant amount of expertise, and honestly it is a matter of time until it gets smashed with something better.
I wish this was true, because a lot of code in Kubernetes is garbage and their entire architecture of constant reconciliation is quite wasteful, but at this point is has essentially won in being supported by vendors. It'll be a long time before something can really push it away. Which sucks, because nobody wants (operational complexity) or can run it outside an enterprise environment, at least not in a way where you can expect a majority of helm charts to work (insanely high compute requirrements for control/instrumentation plane, requirement for loadbalancer and PVC provisioner implementation)
I really want a unified "let's not think about servers" model for using compute and storage but kubernetes is not it for consumers.
Over-complicated "industry standard" solutions tend to fade away after the hype has died down. Examples that come to mind include SOAP and OpenStack. If you think Kubernetes falls into that category, I think it will go the same way before too long.
The difference between OpenStack and Kubernetes is that one was meant to power alternative private clouds and the other just adds an abstraction layer on top of public clouds so everyone can buy into the lie of being independent from them. In my experience the bare metal code paths in Kubernetes are anywhere from mostly untested over entirely broken to just not supported in the first place.
Here are 26 private cloud Kubernetes users who would disagree: https://www.cncf.io/projects/case-studies/?_sft_cloud_type=p...
It's clear that you have zero experience with Kubernetes on bare metal. Why bother commenting on the topic then?
(This is a bit of an exaggeration but you get the idea.)