Hacker News Reader

Comment by bcantrill | Hacker News Reader

Fourteen years ago (!), I reflected on why software vendors are able to get away with bad behavior[1] -- until they can't, of course. That two of the three database companies I mention are in the dustbin of history should be telling -- and I stand by my most basic conclusion that proprietary software companies cannot outrun open source equivalents in perpetuity. May the long run arrive sooner rather than later!

[1] http://dtrace.org/blogs/bmc/2004/08/28/the-economics-of-soft...

View on HN

Any thoughts on business models for Databases going forward? Some interesting recent developments make DBs a challenging vertical to bootstrap in

a. Proprietary is a non starter, since increasingly large customers will insist on the source being available.

b. Integration / Consulting is no longer viable since the cloud providers will likely offer a hosted / managed solution around a new OSS.

c. Licensing is difficult to enforce if you are giving your code away for free.

Will DBs join file systems and debuggers as pieces of the stack where the only way to build one is to be inside a megaco or as a side project? Am struggling a bit to understand how to monetize infrastructure software going forward. (I have ignored the option of raising $X M to build something and be acquired before shipping to a large customer base).

Disagree on B. There's always opportunity for consulting and integration. Not everyone wants to be in the cloud and not everyone can be in the cloud.

And you know DBs always need tuning.

Heh, that's largely Spolsky's "Smart companies try to commoditize their products’ complements."

https://www.joelonsoftware.com/2002/06/12/strategy-letter-v/

It's perfectly possible to have proprietary software -- for a while. But much like patents expire, the advantage of being first to market also expires. If you want to keep your proprietary stack relevant, then at some point you have to open it up. To keep earning income in the software industry you have to keep innovating, and you have to provide the best support.

Sometimes it's best to start open source.

Open source is a business tool. It's not always the right tool, but in the long run it always is just because if you don't make your software open, then your competition will.

So where is the open source PL/SQL development environment, with graphical debugger, data modelling tools and distributed transactions support?

I don't need an IDE. A debugger would be nice, and I have my techniques for debugging PostgreSQL (which has PL PG SQL). Distributed is something being worked, I think. PG serves my needs and is a solid platform.

I'm asking because I dont know, but have you been involved in cases where the OS was free but the db was 40 grand/cpu? Is the os free as an add on from the db vendor?

I can't believe it was that long ago. It feels like yesterday.

I posted the blogs.oracle.com link to your post -- it's very satisfying that it's still up there.

I don't think this is true with SaaS.

You still run it in AWS, Azure, Google.

To an extent though, it is! All of those vendors (some more than others) are pretty serious about Kubernetes, and they're a tiny fraction of the list[1].

I keep hearing that AWS Lambda has won serverless (Functions as a Service), but as a fellow K8S-bettor who abhors vendor lock-in and follows the news from CNCF friends, I'm not sold on that.

Azure has Brigade, Pivotal (Cloud Foundry) has Riff to compete with that, and I'd guess there are other solutions in the same vein that I haven't heard of; well both of these solutions at least are portable to anywhere you run your Kubernetes clusters.

As long as you don't leverage the other vendor SaaS APIs too heavily and keep as much of your control plane inside of Kubernetes as possible, you can always take the solutions you build there with you anywhere.

Those are open-source solutions built on SaaS platforms, and the fact that things like those exist... is pretty much the only reason I'd consider those platforms? Even potentially say that I'd prefer them over AWS (except that we're an AWS centered shop, where you can also run Kubernetes if it's your fancy, and that we do all of those lock-in things...)

[1]: https://docs.google.com/spreadsheets/d/1LxSqBzjOxfGx3cmtZ4Eb...

The 12 open source serverless projects that can be installed on Kubernetes:

https://landscape.cncf.io/grouping=landscape&landscape=insta...

Reply on news.ycombinator.com