Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Intel Pulls Out of OpenStack Effort It Founded with Rackspace

fortune.com

Post reply on HN
Print view View original

71–80 of 118 posts

Re: Intel Pulls Out of OpenStack Effort It Founded with Rackspace

#71

Post by WorldMaker »

jacques_chester wrote: post #67
WorldMaker wrote: post #61

Earlier quoted context omitted.

This seems to get back to an interesting discussion I had with someone early in the OpenStack efforts: as a developer OpenStack doesn't directly interest me because I don't care about infrastructure. Where OpenStack had a possibility to win was to provide options for infrastructure agnosticism: if I can build an app that runs "unmodified" on any OpenStack-based infrastructure, that has a possibility to save me potent…

> if I can build an app that runs "unmodified" on any OpenStack-based infrastructure, that has a possibility to save me potential time and money from having to port apps to/from/between AWS, Azure, Google Cloud, et al (assuming of course that it enough clouds actually adopt). OpenStack is at the same layer as AWS, GCP or Azure. It's not an abstraction over them -- that's a PaaS. I'm not sure OpenStack was ever seriou…

> I'm not sure OpenStack was ever seriously intended to work at that level.

There was a hint at some possibility that it might, especially in some of the hype circles early on. It is of course unfair to blame OpenStack for never actually fulfilling that dream, when that was not its intention per se, but that is part of why today people continue to compare what OpenStack promised versus what containers delivered.

Re: Intel Pulls Out of OpenStack Effort It Founded with Rackspace

#72

Post by mattbee »

foobiekr wrote: post #59

OpenStack really has a bunch of problems. The big one is that OpenStack is about building and operating your own Rackspace; this is not something IT organizations can even hire for let alone carry off. The idea that they could/would/should was purely aspirational. The others are that it's basically a mess - no two OS deployments will look the same - and that it has neither an operational advantage nor too much of a p…

As a small competitor to Racksapce in 2010, designing a replacement for our (then) 8-year old VM structure, Bytemark bet against Openstack from the 1st announcement and spent years building something the-same-but-different.

It wasn't _just_ that Openstack was unreleased and talking about foundations and making grand promises without any production-grade code. It was just that we could never work out which version Rackspace was actually running for their actual customers, and the details of how to actually _deploy_ it and make it work seemed enormously fiddly compared to VMWare, which was where they were taking aim. I expected them to say "now running on Openstack, 100%!"

Also the design seemed to mirror AWS, as if the only answer to their dominance was to ... copy every facility they were producing exactly?

We had a more definite vision of where VMs should go, and thought it was a bad plan to aim at "Amazon, but smaller!". In particular we really really wanted live migrations to be a part of our platform - where we really really cared about uptime of individual VMs, and wanted people to be able to upgrade them on the fly. Plus, y'know, for a hosting company, being in control of, and having opinions of our hosting platform was what people paid us for!

So we designed our in-house platform BigV instead (now Bytemark Cloud Servers) -> https://blog.bytemark.co.uk/wp-content/uploads/2012/12/Desig... [pdf] And even though we (re)invented an NBD server to make all the live migration stuff work https://github.com/BytemarkHosting/flexnbd-c I believe we've ended up with something that does a small number of things far better.

I see people asking whether Openstack is actually running anywhere very public, and have read some pained war stories. So it still feels like the right decision, even before this announcement. Plus it seems like live migration in Openstack is still an exotic and difficult option, whereas that's been our standard practice for years.

There are plenty of problems running your own hosting stack, but I like that we can solve our own problems on our own terms, and tie software decisions to hardware or data centre decisions that we make at the same time.

I eye up Kubernetes & Ganeti every few months, but they keep reminding me that if you've got the expertise, and a specific purpose in mind, you can usually build something that fits your purpose more closely.

(though the second half of that sentence will probably be written on my self-carved tombstone)

Re: Intel Pulls Out of OpenStack Effort It Founded with Rackspace

#73

Post by turnip1979 »

I am sad to see these negative events as someone who was in the openstack community. But I think a lot of folks knew such stuff was going to happen.

I think the mistakes made were:

1) the so-called big tent approach 2) too much complexity 3) core projects not listening to end-users, and focusing too much on the plug-in model

It is a shame because there were so many smart and hard working folks involved. It really felt like a community experience.

Re: Intel Pulls Out of OpenStack Effort It Founded with Rackspace

#74

Post by foobiekr »

mattbee wrote: post #72
foobiekr wrote: post #59

OpenStack really has a bunch of problems. The big one is that OpenStack is about building and operating your own Rackspace; this is not something IT organizations can even hire for let alone carry off. The idea that they could/would/should was purely aspirational. The others are that it's basically a mess - no two OS deployments will look the same - and that it has neither an operational advantage nor too much of a p…

As a small competitor to Racksapce in 2010, designing a replacement for our (then) 8-year old VM structure, Bytemark bet against Openstack from the 1st announcement and spent years building something the-same-but-different. It wasn't _just_ that Openstack was unreleased and talking about foundations and making grand promises without any production-grade code. It was just that we could never work out which version Rac…

The use case I described - Skyport's cloud managed secure servers - needs super high quality, self-recovering embedded code that made no assumptions about the network being high quality or reliable.

I think the thing people - and especially the OpenStack guys - don't appreciate is how terrible it is to (a) lose a workload or (b) require someone visit the datacenter (which may actually be a colo in another state or hours drive away). Having to fall back to some sort of terrible insecure management like IPMI or a dedicated mandatory management network (of questionable security), etc. is just not viable.

Systems and infrastructure architectures need to address a few things that really matter - error handling, continuous self monitoring, state compression and linear, systematic self-recovery - and while some OpenStack components handle this (ceph is pretty great except for a few things around access control and security) the whole doesn't handle them well at all. It's not enough to log a message (or worse, just log an exception stack).

Re: Intel Pulls Out of OpenStack Effort It Founded with Rackspace

#75

Post by kordless »

I ran an OpenStack cluster in my house for a few years. The deployment was managed by a bunch of scripts which I wrote and published to help others learn the basics in deploying an OpenStack cluster, sans most of the more complicated networking stuff. Last count, about 25K unique IPs downloaded the images used in those scripts to launch test instances. At one time I held the top 3rd or 4th link on Google for "openstack install".

A few years ago, I leveraged a bunch of methods in OpenStack to build a thing which would launch instances based on Bitcoin payments to certain preloaded addresses. These addresses allowed "templates" to be associated with cluster capabilities and code to be deployed. Payments to that address in Bitcoin would immediately net you an instance, of a certain type, on someone's infrastructure.

The general idea behind this creation was a way to abstract hardware components into a system by which applications could launch themselves and utilize the resources provided in a fair, secure and trustworthy way. It is my belief this model was WAY ahead of it's time, a precursor to the hybrid models we see emerging today, and lead directly to my personal realization that federation of all systems will be a basic requirement for implementing trusted computing in the future. We're going to need it with AI. Not sure how I know that, but there it is, irrationality and all.

Unfortunately, standardizing deployment methodologies doesn't net you federation. Standardization itself doesn't net you trust, unless everyone can agree on the standard, which is out of necessity done in an non-trustworthy way. Votes of a board, for example, aren't represented with fair consensus, unless there's an algorithm behind the votes. Otherwise, votes end up being slightly irrational, because people behind the votes are slightly irrational. Or very irrational, depending on the company they work for.

And yes, federation can be implemented in a trustworthy way by a corporation, such as Google or Amazon, but that requires all people in that federation (a discrete group) agree they will use this thing to implement trust, even though the thing may not actually have trust implemented in a rational way (i.e. by algorithm).

At the end of the day, OpenStack was doomed not because of container tech, or the structure of the board, or who ran the biggest cluster, or because it was overly complicated for simple use cases.

It failed because it did not implement the basic requirement of delivering trusted infrastructure in a scalable and trustworthy way across a broad range of infrastructure, in a wide range of locations, and do so in a way that separates the use of the infras from the irrationality of humans running the infra.

Until something does that, and does it well, we're stuck with Google, Amazon and other provider's solutions. This is also a good rationalization for the continued increase of cloud services by companies and the continued emergence of hybrid models in the future.

Re: Intel Pulls Out of OpenStack Effort It Founded with Rackspace

#76

Post by specialp »

kefka wrote: post #46
SEJeff wrote: post #30

Earlier quoted context omitted.

Do you work at eBay, because they run kubernetes ontop of openstack, which sounds a bit awful if you really think about it all.

Not really. Kubernetes is stupidly unstable. When I was testing it a while back, the master processes kept dying for no good reason and leaving half-on processes with no networking. And then you get to restart everything. Yay. Whereas if Kubernetes is on something else (OpenStack/Mesos) you at least have control to restart Kubernetes when it shits its brick.

The "a while back" qualifier here is important. Both Mesos and Kubernetes are relatively new projects in a rapidly moving space. FWIW we are running Kubernetes 1.4x with multiple redundant masters in production with no problems. We are also running Mesos/Marathon internally. These projects did have major issues early on but now they are becoming much more robust.

For control we are using CoreOS with kube-aws. It actually works quite nicely as we can scale and upgrade running clusters. The only issue with it right now is upgrading and managing etcd.

I think the whole issue with OpenStack is the overambitious goal to be a one stop from bare metal servers and switches to applications. Kubernetes with Flannel or Calico can take care of the software level networking and provisioning, while there are a lot of other ways to provision VMs for running Mesos or Kubernetes depending on where you are deploying.

Re: Intel Pulls Out of OpenStack Effort It Founded with Rackspace

#77

Post by yeukhon »

WorldMaker wrote: post #70
yeukhon wrote: post #68

Earlier quoted context omitted.

> if I can build an app that runs "unmodified" on any OpenStack-based infrastructure, that has a possibility to save me potential time and money from having to port apps to/from/between AWS, Azure, Google Cloud, et al (assuming of course that it enough clouds actually adopt). > There are ways now to build container clusters that you can ship in parallel to AWS and Azure with very little code difference. OpenStack is…

Yes, it is an unfair comparison. I was attempting to explain in my comment why I consider it a valid unfair comparison, because it is a matter of perspective. As a software developer, do I ever want a private cloud? No. Does my employer? Maybe. Is it my job to tell them how to invest their infrastructure dollars? Quite possibly no, because software development and infrastructure are typically held at arms length. But…

Okay, I was reading it as defending why container would solve what OpenStack was set out to solve, which is the proposition I read from the OP I was replying to.

Of course, I would advise against running a private cloud unless there is a dedicated team of at least a dozen or so. I applaud Digital Ocean for able to survive and make good business from their private cloud. As a developer I totally agree I just want my code to be deployed and that all the appendices are deployed and configured.

Re: Intel Pulls Out of OpenStack Effort It Founded with Rackspace

#78

Post by AndyNemmity »

justinhensley wrote: post #62
AndyNemmity wrote: post #21

Earlier quoted context omitted.

I just moved into an openstack development job, and I'm struggling. So many technologies and complexities you have to keep in mind, even within a single project. I'm hoping it comes together soon. I have not provided a ton of value yet just sitting here learning, and breaking so many things.

It is a lot to ingest. Can you elaborate on the OpenStack development you are doing? OpenStack refers to a number of different projects (Nova, Neutron, Horizon, etc). I would advise getting a high level understanding of how Nova, Neutron, and Cinder/Swift interoperate. Then drill in to Neutron. As with every other cloud platform I've worked with, most of the IaaS complexity lies in Networking, in my opinion. Neutron,…

I am doing Monasca development within an Openstack cloud running on Kubernetes and built with helm.

So for example a development environment I use is a docker-compose environment building Monasca. That environment is a work in progress, so half the time I'm fighting that environment just to make a proper change.

Then from there, I don't have Horizon, so I have to use command line apis directly to set everything that I wish to use.

The whole thing has just been a challenge, every part I've had to learn in some respect. The actual code I've changed in Monasca, has been the simplest part so far. Validating it's correct, and testing it has been the far harder challenge.

Monasca also being a bit of a challenge since it's several different repos of services working together, instead of a large one akin to Nova.

I'll get it, but it's been a struggle so far.

Re: Intel Pulls Out of OpenStack Effort It Founded with Rackspace

#79

Post by AndyNemmity »

jasode wrote: post #33
iamthepieman wrote: post #3

the article mentions other big players pulling out or cutting jobs/funding related to OpenStack in recent months in addition to this latest move by Intel. Is this related to some endemic problem with the technology, big egos within the project, dominance by another outside player or what? Cursory searching doesn't reveal any "big thing". Maybe it's just dying a death by a thousand cuts.

>Is this related to some endemic problem with the technology, I believe the decline of OpenStack is inherent in the type of technology. I don't believe that type of "infrastructure plumbing" software lends itself to high-speed high-quality innovation in the "open source" model. Some reasons: 1) it's boring infrastructure software: Because it is software for "data center" infrastructure, it's not glamorous and interes…

"If the company does decide to run OpenStack, it's likely they won't contribute back any source code. (Same as Linux situation where most users of it don't contribute source code.) This removes another subset of potential contributors."

From my perspective, companies that decide to run Openstack do contribute code back, but the process is so long and arduous to get code accepted that it's not hard to see how groups would just give up and just merge upstream to their own.

Core reviewers have to accept your code, and they are on no timeline to do so. Months will go by without a review.

Re: Intel Pulls Out of OpenStack Effort It Founded with Rackspace

#80

Post by xemdetia »

kashyapc wrote: post #35
xemdetia wrote: post #28

From someone not very invested in the OpenStack process it feels like all the momentum of the recent push of the project just withered away all at once. I am not sure if the container crowd ate it's lunch, where containerized appliances just solve the problem of openstack better.

[Disclosure: I spend a (not too large) _portion_ of my time on the OpenStack Compute project, so adjust my comments for bias.] Please. You seem to be making the same mistake, that many make on blogs and news websites, of exclaiming "containers are the future!" while both VMs and containers have their place and use-cases. I'd rather suggest you read the below. (Since I'm linking to a blog, quick background about the a…

I am not really claiming containers are the future, but from someone who is actively working on a network appliance I have people asking me at least twice a month about containers to put my product in (from people who are already managing 10k+ systems and wanting to simplify deployment). I have heard zero buzz about OpenStack except during a SVP drive by about a year ago.

Most of the customer sites I interact with on the high end just end up baking their own system images tied to a SSO solution and just manage it with VMWare/Ansible/something homegrown. On the low end people just don't deal with OpenStack at all and just do it the hard way with a VMWare or KVM style solution. For either side of that equation containerization is a faster value-add.

From the other side what people ask me about cloud integration is more feature-driven of particular clouds than anything else. I'm just not seeing the momentum in the field.

Post reply on HN