Goodbye, OpenStack | Hacker News Reader

Goodbye, OpenStack

gigofham.com

116 pointsswills25 comments

So I've been in the OpenStack community for a while [3+ years] (http://stackalytics.com/?metric=commits yes that's me down there in the pie chart); and I'd like to answer any questions anyone has...

Yes the OpenStack community isn't perfect (but I've heard the same thing about other communities, so meh).

Something I've found interesting to read over:

- https://lwn.net/Articles/647524/

Overall, no the community isn't perfect, yes there are issues, yes it burns some people out, but software isn't rainbows and butterflies after all.

> Overall, no the community isn't perfect, yes there are issues, yes it burns some people out, but software isn't rainbows and butterflies after all.

Are people trying to make things better? If so, how?

Software isn't rainbows and butterflies, but it shouldn't be endless floods and caterpillars, either.

Yes, I think they are getting better (at what rate, I can not measure, but forward progress IMHO).

One example; the project/group I work in (https://wiki.openstack.org/wiki/Oslo) basically is a bunch of well-known people in the software community (author of sqlalchemy included, authors of python books, myself...). Libraries I (and others) work on are trying to help solve the tech. debt of the various projects (good practices and all that). So I think this has helped make things better...

Obviously progression is relative and certain projects in OpenStack move at different rates (depending on complexity, vendor involvement, or lack of vendor involvement, number and quality of developers doing reviews/commits...).

There's always work to be done (and yes I know the operators pain, I work at yahoo! and sit right next to the operators running our OpenStack based clouds, vm and baremetal...); things aren't always easy but then what software really is in the end...

OpenStack isn't perfect, for sure. Huge community, big corporate players, huge code base. I do feel the barrier to entry into the community is very high; using the stack is not a small undertaking, particularly in a production capacity. The APIs are mess, but I have high hopes that the new OpenStack SDK will bring some ease of use to the stack that boto has done for AWS.

Overall, even though I feel like a reasonably competent coder with a lot of experience as a user, I don't feel like I could ever contribute to the project in a meaningful way - and I don't think I want to.

Working at an Openstack cofounder, I do have a lot of code deployed into an OpenStack cloud, and have written a lot of automation around the stack and APIs. I still have high hopes for the platform.

As someone who's only been on the outside of OpenStack's community, but interested and learning about it a little here and there, could someone give a brief overview of major players and why it seems a number of individuals involved in the OpenStack community have burned out?

Is it really as bad as this post suggests? Many technologies become heavily burdened by too much (non-development) investment and companies doing a land grab on a hot new product; is that the case here?

Redhat, IBM, Rackspace, HP are pretty major players (and there are users like HP, Bloomberg & IBM research). You can find them https://www.openstack.org/foundation/.

While I don't have any first hand experience contributing to OpenStack (but complained quite a lot back in the old days as user), most of the developers are paid. Obviously OpenStack started out with NASA and (then) Rackspace, so it totally makes sense folks there held the manager/committer/BDFL titles and drove the decisions.

I do agree many of the APIs are awkaward and the Python code is not very Pythonic (classic example: why are attributes with "-" exist?? as user I have to use getattr(object, "OS-XXXX") to retrieve the value). Oh my god.

I hope https://wiki.openstack.org/wiki/API_Working_Group (which I'm pretty sure is active @ https://wiki.openstack.org/wiki/Meetings/API-WG) will make this better in the future. As always feel free to show up :)

I'm not familiar with both OpenStack and Python so I might be totally missing the point but JavaScript has a similar problem with DOM attributes and CSS rules. It accesses them either as strings or as camel cased object properties. Example with jQuery

    el.css({"background-color": "red"})
    el.css({backgroundColor: "red"})

Can Python do the same?

>> Is it really as bad as this post suggests?

That's very hard to quantify, one of the projects (really a library) I work on I'm pretty proud of and I think it's actually somewhat innovative: http://docs.openstack.org/developer/taskflow/ (obviously I'm biased).

Observation: The OpenStack board is controlled by board seats which are mostly held by people employed with large publicly traded companies. At one point I calculated these companies to have a combined market cap of ~US$1/2 trillion.

Blaming Statement: Make no mistake, the companies who control OpenStack are "in it to win it". That's probably not good for innovation or optimization.

Reality: Everything above is a moot point because Docker. Or more specifically, everything above is a moot point because LXC containers. Containers are a better solution layer for developers than instances. Instances are still a good solution layer for providers, but those guys are few and far between, and their margins are thin. No clue what that means for OpenStack, but it can't be good.

Disclaimer: I've helped ~25K people install OpenStack with my scripts. I'm biased and am spending most of my time now with containerization technologies.

>> All of that is a moot point now because Docker.

What made you think OpenStack wasn't into containers? :)

- https://wiki.openstack.org/wiki/Magnum

- https://www.openstack.org/blog/2015/07/google-bringing-conta...

- http://adrianotto.com/2015/01/announcing-magnum-caas-for-ope...

I know it's a common misconception but OpenStack is not just virtual machines.

It's a stack and many projects with various names do various things, from compute APIs [nova], volume storage [cinder], networking [neutron] to baremetal [ironic] to orchestration [heat] to ...

> Everything above is a moot point because Docker.

OpenStack does more than containerise.

Docker Inc. are racing up the stack to catch juicy value, but so is everyone else.

Red Hat, Pivotal[0] and IBM are already at the top of the stack with OpenShift and Cloud Foundry, so it's getting crowded.

But in the middle there's an awful lot of mechanism required to run an IaaS and OpenStack is the only serious opensource contender around.

[0] disclaimer: I work in Pivotal Labs.

'the only serious opensource contender around', have you heard of Apache CloudStack? It has seen some serious usage: http://cloudstack.apache.org/users.html

I haven't, thanks for the tip.

Racing up the stack doesn't imply you are dragging your ass up the same stack down below. There are some serious contenders in the container space that simply don't require anything below them but bare metal. They don't need virtualization, and microservices practices are being investigated in every company I'm talking to ATM. OpenStack can do whatever the fuck it wants to, but it's not going to be enough to capture massive market share with consumer coders. Like I said up top, OpenStack is destined to play out it's future in a low margin vertical, who's margins are getting lower by the day.

Docker and Openstack are working at very different levels. Kubernetes is a much closer competitor to Openstack.

Docker is not great for multi-tenant environments.

But Intel is trying to solve this with Clear Containers.

I don't understand why this is on the front page. Take a look at his contributions: http://stackalytics.com/report/users/carlp http://stackalytics.com/report/users/carl.perry@dreamhost.co...

I've met Carl Perry several times at a couple of OpenStack summits and mid-cycle operator's meetups. He's a nice guy, and seems like he's pretty smart, but he's got an opinion about everything and likes to hear himself talk. The last time we were in the same room was when we were in the same working group. I don't remember what our topic was, but all I remember was Carl completely derailing any kind of discussion we had in the room.

I've been part of the OpenStack community for about two years (working for an OpenStack operator) and I've had nothing but good experiences. Developers that I've met seemed genuinely interested in hearing our pain points and other operators are interested in sharing their experiences running OpenStack and learning from our experiences.

It's really not that hard to contribute to OpenStack, you just have to play by the community rules, learn to listen a little, and don't piss people off with arrogance. Oh, and about the price of attending a conference, $600 for a full access pass. Is that a lot?

What a nice project, comes across as something from the government..

"Anonymous sources in the Administration say the author was a junior staff member with no access to classified material who was fired due to poor performance"

It's easy to throw rocks when you post as anon.

Shameful.

Yeah when too many players with overlapping vested interests join in, it will be a disaster. Can you believe VMWare is part of the board Ha ! Ha!

Why not move to other opensource projects like opennebula.org ? By the way I work for Megam - https://www.megam.io another opensource PaaS project https://github.com/megamsys. Let me know if you are interested in any of those.

That other big opensource project (the one called linux) I think would disagree with you, its management style might be different (more of a hierarchy with linus at the top) but saying 'when too many players with overlapping vested interests join in, it will be a disaster' seems disingenuous IMHO.

Know what outweighs all of those problems and makes it worth it over even the medium term? It's open. You don't get tied to a provider who has your business at their mercy.

This reads to me like "massive software project is massive and I can't influence it in the ways I want to".

Much like your ability to influence "Windows" is quite limited, your ability to influence "Openstack" is limited. Pick a neighborhood and you'll likely be successful if you're doing a good job. Pick a continent and you might as well ask the Sun nicely to stop giving off radiation.

Reply on news.ycombinator.com