But at the end of the day any compromising on open-source ideals is going to piss someone off - heck even just choosing one OS license over another is going to piss people off. It doesn't matter. In business you need to have a thick skin to stick to your conviction about what is best for both customers and the company - whatever that may be.
https://blogs.gnome.org/bolsh/2010/07/19/rotten-to-the-open-...
A purely services based model is difficult, but so is open core. With open core you compete against yourself. Look no further than Nginx as an example and how lousy it is to do things like say a http healthcheck of HEAD /status. In haproxy it is trivial, it can be done in nginx plus, but not the oss version (at least it wasn't supported in any of the versions I tried a year ago sans the commercial version). When you compete against yourself you get put in a difficult situation "Does this valuable user contribution to my OSS project compete with my commercial fork?" and that is just all around a bad situation to be in for the community and for the company.
And @yasserf's reply was fantastic, too. Sounds like they are more providing premium hosted services than anything else, which I think is definitely a good way to go.
deepstreamHub literally runs the deepstream.io servers to provide all of the fundamental functionality. All of the client sdks are also Apache 2 based which means that they can be changed in whatever way suits people most, and written in any language.
We also have some really interesting forks by community members that have diverged a bit from how we do things, and has resulted in collaborative forums and in person meetings as well as even talks at meetups about the pros and cons which resulted in the divergence.
The non open aspects are all around the extra services ontop. These all work by hooking into deepstreams plugin architecture, meaning that as stated in following comments, open source plugins are encouraged and are documented in as much detail to help people get started. Offering all the non-core/multi-tenancy/usage monitoring/gui components aspects as open source is in that sense extremely counter intuitive, since we will instead be providing very specific functionality for the purpose of creating another PaaS system. That won't be stopping us from releasing detailed articles/give talks about how things work and run under the hood on AWS and the challenges we met hosting at large scale and across regions, it just means we won't be providing the actual code.
In terms of auto-failover, replication, etc, it would be great if you look at the current open-source aspects on deepstream.io that we have in place for that. deepstream is built with availability and clustering at its core, and its truly because of the involvement of the open-source community that we have managed to make it run in multiple production environments to date.
Would you be willing to call it "open source with premium hosted services" or something rather than "open core"? I feel like that is more accurate and would be better for marketing, and avoid the whole "open core / crippleware" debacle (see the other comments about InfluxDB). Just my 2 cents!
Cheers! (Thanks for the reply)
It isn't the right approach, and we ( http://gun.js.org ) aren't going to take it. More than happy to be the only one left standing, devoted to true open values - but I hope others join us.
major humble-bragging. In bad taste to do that in an announcement for a "competitor"?
Here is my thoughts/arguments around it, dealing with the RethinkDB and Parse shut down as well: https://hackernoon.com/the-implications-of-rethinkdb-and-par...
Plus, was invited to speak about it on the Changelog Podcast: https://changelog.com/podcast/236
Please stay truly open :) that way we can still compete. ;) Or else we'll be the only company that is fully Open Source (replication, auto failover, etc.)!