[gregmac]

> If you're down, my knowledge access is down.

Even for companies that want to cloud host, there's a concern about if something happens to Shelf itself (eg: the service suffers a major multi-day outage or goes out of business). There needs to be a contingency plan (which is probably some form of self-host).

If usage of a product like this is successful, it both becomes a critical operational system and represents thousands of hours of time invested populating it. There needs to be assurance that investment is not lost and the business is not 'down' in case something happens. I work at an SMB and we use quite a few cloud-hosted services, but whenever we do we discuss the balance of how critical a service it is and what our contingency plan would be. Sometimes it's switching to a competitor (and maybe importing history), sometimes it's pulling the open source version and hosting ourselves.

The product looks great, but all I can see is "Data backup and recovery" as a feature. I don't know what that means or what happens if shelf.io disappears, so that's a huge red flag for me, unfortunately.

[tfjaeckel]

Thanks for sharing your thoughts on this topic. It's a challenge for all SaaS products I believe but yes, the more business critical the product the more important to address.

Apart from self-host and open source, do you have other suggestions on how to address this on a product level? Would you feel comfortable with structured exports for example?

[grantlmiller]

Check out www.enterpriseready.io to get a deep understanding of the standard features that enterprises require (ie SSO, RBAC, audit logs, on-prem deployment, reporting etc). Happy to talk through any of the features, I contributed about 80% of the content. We also built Replicated.com which powers the on-prem delivery for other applications like Travis CI, npm, CodeClimate, Circle CI and a bunch of others. We think about these problems all the time and how we can best help solve them. Happy to share what we've learned there as well.

[tfjaeckel]

Enterpriseready is a good site, agreed. I'll also check Replicated to see how you support on-prem and would gladly take your offer to hear more about your learnings.

[gregmac]

Structured exports are certainly better than nothing, but it doesn't address having to find and import to another system (which won't have all the same features) and retrain all staff.

Honestly this is a hard problem and something of a catch-22: the more useful the product is, the more it gets used, so the more critical it becomes, and as a result the pain goes higher.

A long track record of business, knowing if the company is profitable vs burning through VC cash, and some type of SLA and guarantee of minimum shut down notification time could all help lower risk. I don't really have any other product suggestions, unfortunately. My feeling is this is not really a technical problem and so probably doesn't have a purely technical solution.