Nebula is shutting down | Hacker News Reader

Nebula is shutting down

nebula.com

85 pointsjsmthrowaway53 comments

I met Chris Kemp two years ago along with my ACIO at a midwest compute conference. We explained that between Piston and Nebula we chose Piston. When Chris asked why, I explained that it was due to price.

He replied "You mean to tell me the only reason you didn't buy my product, is because of the price?!?!"

To which I replied "That's the only reason I don't drive a Lambo."

Even with a heavy discount, Nebula was too expensive. With that being said, I am equally worried about Piston's future. With the two OpenStack creators moving on from their first OpenStack entrepreneurial efforts, it definitely indicates a new phase of this OpenStack thing.

The details on the shutdown are pretty spare. While the cloud market is perhaps still a bit immature, there are obviously many companies, both venture-backed and not, who are making plenty of money today. With a client list like they advertise, it makes me wonder how they couldn't trim down a bit to become revenue neutral at the very least, or trim their run rate down and raise another round to stick it out another few years. Something smells fishy.

Yep.

https://www.nebula.com/customers/

Look at that huge list of big name clients. Either those clients were using Nebula for a very, very small operation or Nebula wasn't pricing their products accordingly. I mean, Lockheed alone should have been enough to have basically unlimited money.

There's something else here that we haven't been told.

I vaguely recall hearing that Nebula's OSCON party seemed to operate on an unlimited money budget, but I missed that year, and the subsequent year they were notoriously absent.

It used to amuse me that training companies used one person taking one class once as enough excuse to advertise "Microsoft approved training!"

Someone spending training budget on something does require approval, but there was (at the time) almost no oversight. I wonder if this is what's going on here.

They're probably mostly PoCs.

It's pretty common for companies to buy a small license just to experiment in the lab. At the end of the year you have some extra discretionary dollars, why not make a small purchase with startup that came by a few months ago?

It's not good for NSA if companies/people have private clouds.

April fools? It seems like an awfully poor way to do business for such a company to just go under without any advance notice. But then what are the company's customers going to do? Go somewhere else?

Well if it's just OpenStack, or if it's software built on top of OpenStack, it's fully possible to migrate to another OpenStack vendor. Then again, that supposes the product is open source.

That also pre-supposes that openstack implementations are interoperable.

Nebula was coupled with hardware that included a bundled Arista switch [1]. The hardware provided additional security features. (TPM etc.) So it might be hard to switch to another Openstack vendor from a stack so tightly bound to hardware.

1. http://www.theregister.co.uk/Print/2011/07/27/nebula_opensta...

https://twitter.com/bcantrill/status/552551004114534403

> April fools?

It's possible, but if it were I wouldn't expect them to tweet that it's true[1].

[1] https://twitter.com/NebulaInc/status/583388942294097920

April 1 is also the beginning of a new business quarter. Could very well be that they were looking at the quarterly finances, and decided not to continue running a sinking ship.

Most definitely not April fools, just terrible timing. Source: I was Nebula's CTO.

That's unfortunate. I just saw you at SLAC a few weeks ago. I was literally logging into the nebula about 10 minutes before I read this, helping out my boss with something. I told him, but we weren't sure if this was April Fools or not.

I'm very sorry to heard that, but still I hope yourself and your team continue working with cloud computing, ideally OpenStack ;)

I'm really disappointed they decided to shutdown everything so suddenly with no plans to keep support going for at least another 3 months. What kind of management team gets a company to a point they have to shut it down in a single day and tell everybody to pack and go home?

For posterity (and to avoid entering similar risky situation in the future), here is their management team:

- Gordon Stitt (@gordonstitt)

- Chris C. Kemp (@kemp)

- Devin Carlen (@devcamcar)

- Vish Ishaya (@vish)

- Herb Schneider

- Huy Nguyen

And additional board members (mostly VC supervisors):

- Peter Bell

- Mark Leslie

- Ted Schlein

- Louis Toth

I know this is a vapid comment but the video on their website showcasing the Nebula Cloud Controller is slick.

They did make great videos!

The voice of Patrick Stewart at the start of the Nebula One launch video still gives me shivers: https://www.youtube.com/watch?v=r4rsEjsomFI

And don't forget back in 2011 with the launch of the company at OSCON with the who's who intro video before the presentation including hardware pr0n https://www.youtube.com/watch?v=3oydzMfFYWM

I also couldn't help, but think that http://opennebula.org/ will be happy for the end of the namespace collision.

Hubris videos aside, Nebula.com carried more than their share of the weight in the first 3 years of OpenStack, and made me believe that OpenCompute wasn't just Facebook. Thank you and best wishes to all who brought NASA Nebula to open source and made OpenStack a reality!

"including hardware pr0n" ))) NSFW (sound wise)

I have no idea what paying for Patrick Stewart to do a voiceover costs, but I can't help but feel it may have been a contributory factor in their demise. (Not, perhaps, directly, but it may be indicative of the sort of mentality that suggests frugality isn't a high priority...)

Can I have the domain name?

I reckon the domain name will be one of the many assets sold off to recover any losses as a result of the shut down. But having said that, if you don't ask you don't get!

April Fool? Please don't do this sort of public announcement on Apr,1.

I worked on Openstack for about a year and moved on something else, it was not quite ready then, maybe still not there yet?

My last employer wanted to look at OpenStack as a drop-in cheaper replacement for VMWare vSphere - and unfortunately, it wasn't that simple.

April 1st

This is a pretty poor April Fools.

Sounds like an April Fools joke...

I agree with Simon Wardley here [0] - OpenStack is a joke (from a business perspective) and all companies trying to build businesses on it will perish as Amazon builds a dominant monopoly.

EDIT: Not companies that use OpenStack, companies that attempt to sell "enterprise clouds" based on OpenStack (like Nebula).

[0] - https://twitter.com/swardley/status/499952875813224448

I don't know how to parse your comment. If you mean that companies using openstack as their core intellectual property, and providing add-ons and ancillary systems for it will perish, I think that may be so. If instead you suggest that any company using openstack in their business will perish, I think that's false. At the highest level, AWS lacks a private cloud on-site solution, for all but the CIA, AFAIK. Many companies can/will/are us[e|ing] openstack as part of their business to provide a private cloud. Arguably they'd be better off w/ native xen, lxc, docker, etc, but they seem to be making it work.

The former interpretation is what I meant and I have updated my comment to reflect it.

The future AWS is driving towards is where they are the power generation company that nobody has onsite either (except for very rare circumstances).

You might keep some temporary capacity / backups locally, but you'll depend on the utility to be there for the vast majority of things.

The market will see if that's the future or not...

That's the "Big Switch" theory. It's one possible future although my sense is that there's considerable skepticism about it from both the perspective of where the economical scale points sit and with respect to governance/control.

> The future AWS is driving towards is where they are the power generation company that nobody has onsite eithe

Don't you mean "a power generation company." I take it you live some place that does not give you choice.

You don't chose who provides the electrons coming over your poles in the VAST Majority of locations (some folks run substations to multiple grids, but even then, are talking to the whole other grid, not a single power company).

Deregulated power is about where you send your bill, not who's producing the electrons.

I'm sure if someone paid Amazon enough money they would consider building out their very own GovCloud. However, I suspect that the amount of money to do so would be so high that the only way it'd make financial sense would be when you're as hopelessly dysfunctional and leaking money out every pore like the Pentagon. I have a customer on track to be one of the largest users of AWS in the world within several years and yet they're not asking for their own datacenter despite the millions and millions that'll be spent.

http://aws.amazon.com/govcloud-us/

So, are you suggesting that if you want computing infrastructure your one and only option will be to use AWS? Nothing against AWS--I use it for some things personally--but that suggests a future where the 1990s Microsoft quasi-monopoly is nothing in comparison.

Google is putting substantial resources behind Cloud Platform: https://cloud.google.com/

But in general, yes, there are serious capital costs and economies of scale with infrastructure, so there won't be a lot of small competitors in this space.

Do you get to chose between electricity providers?

I wonder how big AWS gets before it becomes a regulated utility and 'critical infrastructure' (they hate that question, btw).

> Do you get to chose between electricity providers?

I know this is not your point, but in Germany, for example, you do get to choose. They've created competition between electricity providers.

> Do you get to chose between electricity providers?

Texas also. Compare pricing here: http://PowerToChoose.org

You don't actually chose between providers, you get the same electricity. You simply choose who gets your cash from all the generators on the local grid.

It's not like they run new wires to your house or that your house gets a special charge that your neighbors don't.

> Do you get to chose between electricity providers?

Yes, here in Pennsylvania. Heck, the providers market directly to me, and I've actually made a choice.

You chose who you pay, but do they run separate cabling to your house? Or are you buying power from 'the grid'?

I hope that your post is an April Fools joke. We've been running our company on OpenStack for the last year (longer if you count Rackspace's OpenStack-based public cloud) and it's been awesome. I gave a presentation on it last night at the Seattle DevOps meetup:

https://speakerdeck.com/chrissnell/so-long-public-cloud

EDIT: In response to your edit (companies that sell OpenStack), I also disagree! OpenStack is still a bit of a bear to maintain and upgrade. It requires a lot of specialized knowledge and many companies that might leverage it may not because of the amount of time required to set up and maintain the infrastructure. We utilize a managed OpenStack cloud at Rackspace. For various reasons, building a collocated cluster based on Nebula was not evaluated as a hosting option for us but it might have been viable. I think there's a market for easy-to-operate OpenStack and I'm not sure why Nebula failed. I'm guessing (based on some other comments here) that there's more to the story.

Well hey there! That was a great talk!

That is a good presentation!

OpenStack tries to support too many different backends for most components and is a consultants' wet dream when it comes to custom work and patches required that can be readily milked into an expensive and lengthy deployment, as well as expensive post-deployment maintainence and upgrades. It's only that at a significant (Netflix) scale can the costs of DIY be cheaper than just paying the AWS tax. (Last I heard, Yahoo employed a team dedicated to maintaining OpenStack. Also, I hear from others that this is the norm, not the exception. We looked at OpenStack at Stanford, but there are other IaaS clouds that have better opinions about architectural choices (only supporting say Xen and Vmware because they're production proven) and less code.)

Technically, there is an OpenStack installation at Stanford (at SLAC). It's mind boggling that SLAC and Stanford hasn't banded together on this yet, especially considering the new-ish "Farm" at SLAC.

SLAC has a larger budget, better staff and processes with narrower use-cases than campus proper. They run a decent IT shop.

Reply on news.ycombinator.com