The IT world’s love-hate relationship with OpenStack | Hacker News Reader

The IT world’s love-hate relationship with OpenStack

gigaom.com

18 pointsiProject15 comments

Having spent a lot of time with OpenStack, CloudStack and a number of other cloud platforms (OnApp, Hexagrid and others) over the past two years I came to a conclusion about OpenStack. It's like a bag of dough. Which is great if you're trying to make bread. But if what you want to make is a sandwich, it's not remotely there.

What I mean by that is, OpenStack seems very immature despite how much effort has gone into it. It's still missing a lot of features that, if you're trying to turn it into a commercial cloud platform, are necessary.

CloudStack, to me, seems more complete. I've been working with it for a couple months now and have been very happy with it. It's what we'll be rolling out to production hopefully later this year.

> It's still missing a lot of features that, if you're trying to turn it into a commercial cloud platform, are necessary.

Could you elaborate on this a bit?

One big issue is usage records. Last I checked OpenStack didn't have good API calls for collecting usage data for users/VMs. It was one of those things that was being worked on but wasn't really fully baked yet. CloudStack has a great implementation already there and fully functional.

I see. What else caught your eye?

Honestly it's been ~6 months since I was messing with OpenStack so the other issues I ran into are not on the forefront of my mind right now.

But all of them, from my perspective, were a result of OpenStack just not being fully baked and ready to deploy in a production environment. It needed more development on our part in order to get it to where we'd need it to be compared to CloudStack.

It was always a case of missing features / functionality rather than something being broken / crappy, at least.

Thanks for your input.

I guess it depends on what perspective you're looking for. OpenStack sends usage information to an AMQP queue for consumption. This is generally for use by the provider and not the end-user however.

This is how the currently deployed public clouds collect usage data.

Like I said, it doesn't provide nice simple API calls for collecting usage information. You have to collect it yourself (from the queue, using PSHB or something.) Then you have to then make your own interface for your billing system to then request records it is interested in from this collection.

CloudStack providing simple API calls that allow you to request usage for any time period is much nicer. It can interface more directly with a billing system.

It's just one of the examples of OpenStack seeming to be kind of "halfway there", whereas CloudStack is much closer to being ready to go.

The queue of usage events is an API for collecting usage information, but it sounds like it's not as high a layer as you'd like to see.

There is also the (new) Ceilometer project which sounds like it implements what you're looking for:

https://github.com/stackforge/ceilometer

It is new, so it's probably not as mature as other options, but it has had a lot of activity recently.

There is no doubt that the velocity Amazon ships features is a challenge to both OpenStack and CloudStack.

OTOH, I've been very impressed with HP Cloud (which is OpenStack based). It's true that it isn't yet feature-comparable with Amazon (if you use all the features of Amazon), but for most EC2/S3/CloudFront users then the HP Cloud combination of Compute (Nova)/Storage (Swift)/CDN (Akamai) matches up well.

Following the dev list, and it is clear that OpenStack comes from Operations rather than Software roots. For example, there is a lot of effort and discussion around the command line tools, but the edge cases in the actual web interfaces aren't well documented at all. (I"ve heard that CloudStack tends the other way, but I don't follow that as closely).

I recently ran a survey[1] that showed that over 50% of users have resources with more than one provider. To me that shows that there is room for things like OpenStack.

[1] http://fifthvertex.com/2012/08/07/cloud-tools-survey/

Having looked into and setup test implementations of Cloudstack, Openstack, Opennebula and a few others... I have to say I was not that impressed with the current state of Openstack.

For starters the documentation left much to be desired. An example of this was the implementation of their XCP support. It exists, it's a thing in the stack... but good luck implementing this based off of the documentation! They fail to mention that it isn't really possible without an extreme amount of hacking. In general the documentation leaves a lot to be desired in my opinion, and for a project seeking as much influence on the cloud space as Openstack is, this is a big miss.

Cloudstack has its own issues, but was a reasonably well documented project with a clear implementation path. To note though, is that comparing their XCP/Xenserver compatibility isn't fair since Citrix sponsors this project...

I think one of the problems with Openstack is it is trying to be too many things when the core foundation of the software does not seem to be fully 'defined' yet.

What about someone like Joyent (www.joyent.com) vs OpenStack? They're more of a software player but claim to have a cloud stack that operates more efficiently than others.

You're talking about smartOS I'm assuming? I'm not certain the comparison is apt. With smartOS you're just being given a node platform, with Openstack you have the entire management platform.

Well i think he is referring to Smart DataCenter, which runs on Smart OS etc.

Yep Smart Data Center.

Reply on news.ycombinator.com