I've watched three different teams fail to get vault up and running in any kind of a sustainable way. If they could solve that problem and add a desktop client they could crush 1password in this space. Probably wouldn't hurt that tons of software engineers are absolutely pissed at their moves in the consumer space recently.
As a (paying) Hashicorp customer, my #1 problem is that there doesn't seem to be a lot of talent out there that actually knows how to use these tools and is deeply experienced in them.
And they are all critical pieces of infrastructure. I am very nervous to run a business critical site and store all its secrets in Vault, service discovery in Consul, and containers in Nomad (even though I really like all 3 and they've worked well so far) because we've had a couple of catastrophic cluster crashes where none of us had a clue how to recover without downtime and data loss. (Fortunately they happened in QA but it has delayed our production rollout of this infrastructure)
And I still don't have access to an "expert" – apart from one vendor who seems to know it inside out but is extremely expensive. So there isn't much of a choice.
So then there are two things I can do: become that expert and add that to our business offerings, or abandon the stack since it's too risky to be stuck with it. I don't want to make the mistake I made with Ember.js (nobody really knows it, it's beyond dead outside of a half dozen fortune 500 companies). I don't think that'll happen with Vault etc cuz they really are great products and ideas.
But it is a concern.
Hashicorp must know this, which is why they are betting big on HCP. I'd gladly pay for the uptime of these components to be their problem. And I am paying.
