Hi All,
I was hoping someone could clarify the use cases for the cloud image, particularly in light of the reductions being looked at (like removing python). I ask because I honestly don't know what purpose the cloud image is serving any longer.
The atomic image is squarely targeted at being small, and for running containers. It is somewhat positioned as a CoreOS solution. With that being the case, I'm curious how the cloud image is different and not a repetitive image simply not using the atomic mechanisms.
thanks.
josh
2015-07-10 13:59 GMT+02:00 Josh Boyer jwboyer@fedoraproject.org:
Hi All,
I was hoping someone could clarify the use cases for the cloud image, particularly in light of the reductions being looked at (like removing python). I ask because I honestly don't know what purpose the cloud image is serving any longer.
That's a trade-off * minimize disk usage that comes with a cost on IaaS platforms. * minimize attack surface by removing unnecessary components * base image is a generic one, not everyone needs a python interpreter especially when you deploy a java or ruby apps. * cloud instances provides tooling to customize images at startup (cloud-init in our case) * you can save a customized image to be reused on most IaaS platforms. * Patterns of deployment in the cloud are based on the assumptions of full automation and minimal human interventions.
In my own opinion, base image is a neutral base and one has to customize it for their use. We provide the tooling and a suitable base and it's a much harder task to minimize fedora than adding missing tools.
And that's again, my own opinion, but the day we will provide an easy way to generate customize/upload in IaaS cloud images for end-users (maybe through copr), the cloud story will be complete. Suse has done an awesome job in that direction through kiwi.
As for Python, that's still in discussion, as we need it for dnf and cloud-init. Until we find suitable replacements for these components, it's unlikely to change.
The atomic image is squarely targeted at being small, and for running containers. It is somewhat positioned as a CoreOS solution. With that being the case, I'm curious how the cloud image is different and not a repetitive image simply not using the atomic mechanisms.
One could see the current cloud image as a transitional step between the opinionated approach from Atomic Host and classical linux distro.
Atomic Host concept goes further than just CoreOS, as it could become the next way to envision linux distro including the desktop. Cloud images are the current production, Atomic Host is the Research lab of the cloud SIG.
Regards, H.
thanks.
josh _______________________________________________ cloud mailing list cloud@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/cloud Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct
On Fri, Jul 10, 2015 at 8:18 AM, Haïkel hguemar@fedoraproject.org wrote:
2015-07-10 13:59 GMT+02:00 Josh Boyer jwboyer@fedoraproject.org:
Hi All,
I was hoping someone could clarify the use cases for the cloud image, particularly in light of the reductions being looked at (like removing python). I ask because I honestly don't know what purpose the cloud image is serving any longer.
That's a trade-off
- minimize disk usage that comes with a cost on IaaS platforms.
- minimize attack surface by removing unnecessary components
- base image is a generic one, not everyone needs a python interpreter
especially when you deploy a java or ruby apps.
- cloud instances provides tooling to customize images at startup
(cloud-init in our case)
- you can save a customized image to be reused on most IaaS platforms.
- Patterns of deployment in the cloud are based on the assumptions of full
automation and minimal human interventions.
In my own opinion, base image is a neutral base and one has to customize it for their use. We provide the tooling and a suitable base and it's a much harder task to minimize fedora than adding missing tools.
OK. So your answer to my immediate question is "neutral base that people have to customize". Fair enough. Now, why would someone wish to choose a Fedora cloud image over Ubuntu or CoreOS or any of the other "minimal base that you have to customize" images?
And that's again, my own opinion, but the day we will provide an easy way to generate customize/upload in IaaS cloud images for end-users (maybe through copr), the cloud story will be complete. Suse has done an awesome job in that direction through kiwi.
They've done that with kiwi for a long time. They've had JeOS around for a long time. Which is kind of my follow up point. Why would someone choose Fedora cloud? What makes it compelling?
The atomic image is squarely targeted at being small, and for running containers. It is somewhat positioned as a CoreOS solution. With that being the case, I'm curious how the cloud image is different and not a repetitive image simply not using the atomic mechanisms.
One could see the current cloud image as a transitional step between the opinionated approach from Atomic Host and classical linux distro.
Or, in other words, a minimal linux distro that you download and then have to spend time customizing :).
Atomic Host concept goes further than just CoreOS, as it could become the next way to envision linux distro including the desktop.
Yes, I'm aware that Atomic technology is broader than just cloud.
Cloud images are the current production, Atomic Host is the Research lab of the cloud SIG.
That is somewhat confusing. For the F22 release, I heard much more chatter and excitement around Atomic than I did Cloud images. To the point where I thought the Atomic image _was_ the cloud image for a very long time. Hence my follow up now.
josh
On 07/10/2015 07:28 AM, Josh Boyer wrote:
OK. So your answer to my immediate question is "neutral base that people have to customize". Fair enough. Now, why would someone wish to choose a Fedora cloud image over Ubuntu or CoreOS or any of the other "minimal base that you have to customize" images?
It depends on the use case (which seems like a recursive statement in this thread). ;)
Our customers (disclaimer: I work for Rackspace) usually choose the operating system for cloud instances that they're most familiar with or the ones that mesh will with their organization's strategy. Ubuntu seems to be a popular choice due to the million howto's laying around for installing services on Ubuntu.
When it comes to the ultra-minimal OS choices largely intended for container platforms, like CoreOS, Atomic, or RancherOS, the customer usually has an idea of how they're planning to integrate/automate those operating systems on multiple instances.
Whenever I've spoken with customers about what they want from an OS in a virtual machine, they want it to contain a small package set that lets them run their automation on top of it (i.e. Ansible, Chef, Puppet). Removing Python from that image would be a serious curveball since most people expect to have Python available on any system running yum/dnf.
-- Major Hayden
On 07/10/2015 08:52 AM, Major Hayden wrote:
On 07/10/2015 07:28 AM, Josh Boyer wrote:
OK. So your answer to my immediate question is "neutral base that people have to customize". Fair enough. Now, why would someone wish to choose a Fedora cloud image over Ubuntu or CoreOS or any of the other "minimal base that you have to customize" images?
It depends on the use case (which seems like a recursive statement in this thread). ;)
Our customers (disclaimer: I work for Rackspace) usually choose the operating system for cloud instances that they're most familiar with or the ones that mesh will with their organization's strategy. Ubuntu seems to be a popular choice due to the million howto's laying around for installing services on Ubuntu.
When it comes to the ultra-minimal OS choices largely intended for container platforms, like CoreOS, Atomic, or RancherOS, the customer usually has an idea of how they're planning to integrate/automate those operating systems on multiple instances.
Whenever I've spoken with customers about what they want from an OS in a virtual machine, they want it to contain a small package set that lets them run their automation on top of it (i.e. Ansible, Chef, Puppet). Removing Python from that image would be a serious curveball since most people expect to have Python available on any system running yum/dnf.
Well, it kind of has to be available until we make yum/dnf not need Python anymore, which sounds like *loads* of work.
Personally, I like (and use) the cloud image for two use cases.
1) Openstack development with local virtualization. Having a small-footprint image that has cloud-init is awesome for testing coordination or multinode installations of databases/services.
2) Cloud infrastructure (AWS/Rackspace) a) an OS I'm familiar with b) has great docs/community c) is pretty low-resource/cheap to run d) has up-to-date stuff e) automation I've written for RHEL-derivatives "just works" f) works everywhere I want
On Fri, Jul 10, 2015 at 07:52:29AM -0500, Major Hayden wrote:
Whenever I've spoken with customers about what they want from an OS in a virtual machine, they want it to contain a small package set that lets them run their automation on top of it (i.e. Ansible, Chef, Puppet). Removing Python from that image would be a serious curveball since most people expect to have Python available on any system running yum/dnf.
Yeah, I'm willing to back down on wanting to remove Python.
On 07/10/2015 09:48 AM, Matthew Miller wrote:
On Fri, Jul 10, 2015 at 07:52:29AM -0500, Major Hayden wrote:
Whenever I've spoken with customers about what they want from an OS in a virtual machine, they want it to contain a small package set that lets them run their automation on top of it (i.e. Ansible, Chef, Puppet). Removing Python from that image would be a serious curveball since most people expect to have Python available on any system running yum/dnf.
Yeah, I'm willing to back down on wanting to remove Python.
+1 It'd be a lot of work (rewriting dnf/whatever, maintaining both codebases, etc) for (relatively) little gain.
2015-07-10 15:59 GMT+02:00 Ryan Brown rybrown@redhat.com:
On 07/10/2015 09:48 AM, Matthew Miller wrote:
On Fri, Jul 10, 2015 at 07:52:29AM -0500, Major Hayden wrote:
Whenever I've spoken with customers about what they want from an OS in a virtual machine, they want it to contain a small package set that lets them run their automation on top of it (i.e. Ansible, Chef, Puppet). Removing Python from that image would be a serious curveball since most people expect to have Python available on any system running yum/dnf.
Yeah, I'm willing to back down on wanting to remove Python.
+1 It'd be a lot of work (rewriting dnf/whatever, maintaining both codebases, etc) for (relatively) little gain.
Depends: for end-users, it could mean a smaller bill each month on storage.
But I agree that we're not ready to drop python and it's *unlikely* before a long time. I also agree that it shouldn't be a high/medium priority task.
As for dnf, there are discussion upstream to rewrite it in C, VMWare has also written a drop-in replacement for dnf based on hawkey/librepo that could be considered.
The more problematic component is cloud-init.
-- Ryan Brown / Software Engineer, Openstack / Red Hat, Inc. _______________________________________________ cloud mailing list cloud@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/cloud Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct
On 07/10/2015 10:03 AM, Haïkel wrote:
2015-07-10 15:59 GMT+02:00 Ryan Brown <rybrown@redhat.com mailto:rybrown@redhat.com>:
On 07/10/2015 09:48 AM, Matthew Miller wrote:
On Fri, Jul 10, 2015 at 07:52:29AM -0500, Major Hayden wrote:
Whenever I've spoken with customers about what they want from an OS in a virtual machine, they want it to contain a small package set that lets them run their automation on top of it (i.e. Ansible, Chef, Puppet). Removing Python from that image would be a serious curveball since most people expect to have Python available on any system running yum/dnf.
Yeah, I'm willing to back down on wanting to remove Python.
+1 It'd be a lot of work (rewriting dnf/whatever, maintaining both codebases, etc) for (relatively) little gain. Depends: for end-users, it could mean a smaller bill each month on storage.
Yes, but the engineer-hours to make that transition might better spent making the cloud image (or our docs, or whatever) awesome in other ways.
But I agree that we're not ready to drop python and it's *unlikely* before a long time. I also agree that it shouldn't be a high/medium priority task.
As for dnf, there are discussion upstream to rewrite it in C, VMWare has also written a drop-in replacement for dnf based on hawkey/librepo that could be considered.
The more problematic component is cloud-init.
Indeed. CoreOS has a sorta-ish-done cloud-init written in go, but it's nowhere near feature complete last I checked.
2015-07-10 14:28 GMT+02:00 Josh Boyer jwboyer@fedoraproject.org:
OK. So your answer to my immediate question is "neutral base that people have to customize". Fair enough. Now, why would someone wish to choose a Fedora cloud image over Ubuntu or CoreOS or any of the other "minimal base that you have to customize" images?
We should focus on reliability, availability and predictability, as we have a short-term lifecycle. => we did a great work to fix the first point, kudos to Kushal, Mike and David for that.
predictability is our weak point: * not enough documentation * disruptive changes unannounced (like the ps one recently)
Another weakness is that we don't provide a large panel of application stacks (and versions) unlike Ubuntu. We only provide *one* release of python2, python3, ruby, java, php (though more are available through third-party repo) etc.
That's where the Stack&Env WG work is important for us, as it could become an asset against our other offers.
CoreOS is geared toward innovators and early adopters, our audience which are early adopters and early majority are still not ready to change to the container-based paradigm, as they're already struggling with scalability, automation etc...
They've done that with kiwi for a long time. They've had JeOS around for a long time. Which is kind of my follow up point. Why would someone choose Fedora cloud? What makes it compelling?
We're more aggressive in pushing latest technologies compared to Suse and we do it well. Moreover, we're a good lab to work on supporting the next EL flavors.
A good mix between reliability/modernity
Or, in other words, a minimal linux distro that you download and then have to spend time customizing :).
yes, and here, it's my own opinion, after curating this base, providing tooling/documentation to make customization easier is the key for our success. My peers may disagree with me.
That is somewhat confusing. For the F22 release, I heard much more chatter and excitement around Atomic than I did Cloud images. To the point where I thought the Atomic image _was_ the cloud image for a very long time. Hence my follow up now.
That's understandable, and Atomic/Docker is clearly the exciting side from the cloud lines. We also don't spend enough efforts on marketing our cloud images, and most of ambassadors are not very cloud-aware. I'd rather say that we're more or less starting to be a mature product -though this is a constant effort- , need to focus on polishing and marketing it.
H.
josh _______________________________________________ cloud mailing list cloud@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/cloud Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct
On 07/10/2015 12:59 PM, Josh Boyer wrote:
The atomic image is squarely targeted at being small, and for running containers. It is somewhat positioned as a CoreOS solution. With that being the case, I'm curious how the cloud image is different and not a repetitive image simply not using the atomic mechanisms.
That's sort of a key difference -- atomic == I can't just dnf install things. cloud == I can add on what I want the way I'm used to doing. (e.g., not containerized)
On Fri, Jul 10, 2015 at 11:26 AM, Joe Brockmeier jzb@redhat.com wrote:
On 07/10/2015 12:59 PM, Josh Boyer wrote:
The atomic image is squarely targeted at being small, and for running containers. It is somewhat positioned as a CoreOS solution. With that being the case, I'm curious how the cloud image is different and not a repetitive image simply not using the atomic mechanisms.
That's sort of a key difference -- atomic == I can't just dnf install things. cloud == I can add on what I want the way I'm used to doing. (e.g., not containerized)
Yes, absolutely. However, if that is the only difference then I'm not sure how compelling it is when you compare it to all the other images provided elsewhere that let you do that already. Conversely, atomic is compelling _because_ of the Atomic platform. Atomic has novelty (for now), decent technical advantages, and a lot more marketing behind it.
So what I'm really after is what sets Fedora Cloud apart from every other distro cloud image. What usecases is it better at than {Ubuntu, SuSE, <whatever>}. How should it be positioned so the people want to use it over those, etc. Fedora, in the Cloud space, is still behind in market share compared to the rest of the Linux world. I'm really curious if that can be overcome, or if instead the focus should be on Atomic entirely because it has a better chance.
(To be fair, Workstation and Server also have the same questions to answer in comparison to their peers. However, they have both history and familiarity on their side. Fedora has traditionally been a "desktop" OS and Server can feed off of RHEL which dominates the enterprise space. Cloud doesn't share that luxury.)
josh
On 07/10/2015 04:42 PM, Josh Boyer wrote:
On Fri, Jul 10, 2015 at 11:26 AM, Joe Brockmeier jzb@redhat.com wrote:
On 07/10/2015 12:59 PM, Josh Boyer wrote:
The atomic image is squarely targeted at being small, and for running containers. It is somewhat positioned as a CoreOS solution. With that being the case, I'm curious how the cloud image is different and not a repetitive image simply not using the atomic mechanisms.
That's sort of a key difference -- atomic == I can't just dnf install things. cloud == I can add on what I want the way I'm used to doing. (e.g., not containerized)
Yes, absolutely. However, if that is the only difference then I'm not sure how compelling it is when you compare it to all the other images provided elsewhere that let you do that already. Conversely, atomic is compelling _because_ of the Atomic platform. Atomic has novelty (for now), decent technical advantages, and a lot more marketing behind it.
Well, I mean... it's compelling for us because we want people to have Fedora available $all_the_places, right?
So having only Atomic means we'd basically be saying if you want to do things in the cloud, either do them the "Atomic way" or use another project, right?
So what I'm really after is what sets Fedora Cloud apart from every other distro cloud image. What use cases is it better at than {Ubuntu, SuSE, <whatever>}. How should it be positioned so the people want to use it over those, etc. Fedora, in the Cloud space, is still behind in market share compared to the rest of the Linux world. I'm really curious if that can be overcome, or if instead the focus should be on Atomic entirely because it has a better chance.
I don't think the use case for a generic, modifiable image is going away anytime soon. As far as features go - I'm not sure there's tons of daylight between what Ubuntu and Fedora "do" -- just (as you mention below) there's more in common between two Linux distros on the desktop than not.
Open to suggestions as to what we could do to make Fedora massively more compelling.
I think we'd be sending the wrong message by abandoning the generic cloud image, though.
(To be fair, Workstation and Server also have the same questions to answer in comparison to their peers. However, they have both history and familiarity on their side. Fedora has traditionally been a "desktop" OS and Server can feed off of RHEL which dominates the enterprise space. Cloud doesn't share that luxury.)
On Fri, Jul 10, 2015 at 11:51 AM, Joe Brockmeier jzb@redhat.com wrote:
On 07/10/2015 04:42 PM, Josh Boyer wrote:
On Fri, Jul 10, 2015 at 11:26 AM, Joe Brockmeier jzb@redhat.com wrote:
On 07/10/2015 12:59 PM, Josh Boyer wrote:
The atomic image is squarely targeted at being small, and for running containers. It is somewhat positioned as a CoreOS solution. With that being the case, I'm curious how the cloud image is different and not a repetitive image simply not using the atomic mechanisms.
That's sort of a key difference -- atomic == I can't just dnf install things. cloud == I can add on what I want the way I'm used to doing. (e.g., not containerized)
Yes, absolutely. However, if that is the only difference then I'm not sure how compelling it is when you compare it to all the other images provided elsewhere that let you do that already. Conversely, atomic is compelling _because_ of the Atomic platform. Atomic has novelty (for now), decent technical advantages, and a lot more marketing behind it.
Well, I mean... it's compelling for us because we want people to have Fedora available $all_the_places, right?
Not always. We don't want people to have Fedora on their phones. :)
So having only Atomic means we'd basically be saying if you want to do things in the cloud, either do them the "Atomic way" or use another project, right?
The perception I've seen already indicates we're going that way. If all the hype is around containers these days, even the Fedora cloud download page plays Atomic up. It positions Atomic as the solution for containers and makes no mention of the fact that the base image would work too. It just says it's flexible.
So what I'm really after is what sets Fedora Cloud apart from every other distro cloud image. What use cases is it better at than {Ubuntu, SuSE, <whatever>}. How should it be positioned so the people want to use it over those, etc. Fedora, in the Cloud space, is still behind in market share compared to the rest of the Linux world. I'm really curious if that can be overcome, or if instead the focus should be on Atomic entirely because it has a better chance.
I don't think the use case for a generic, modifiable image is going away anytime soon. As far as features go - I'm not sure there's tons of daylight between what Ubuntu and Fedora "do" -- just (as you mention below) there's more in common between two Linux distros on the desktop than not.
Open to suggestions as to what we could do to make Fedora massively more compelling.
Massively? Probably nothing.
I think we'd be sending the wrong message by abandoning the generic cloud image, though.
Sure, maybe. So instead maybe try sending just as strong of a message for the Cloud image as is done for the Atomic image. This is really primarily a marketing issue. The more I try and figure this all out, the more I think Cloud is being overshadowed by Atomic and keeping them together is detrimental in the long run.
My perspective is going to be much different from someone that lives and breaths cloud on a daily basis. But if I'm confused, there are other people out there wondering the same thing and clearing it up will help more than just me.
josh
On 07/10/2015 05:01 PM, Josh Boyer wrote:
On Fri, Jul 10, 2015 at 11:51 AM, Joe Brockmeier jzb@redhat.com wrote:
On 07/10/2015 04:42 PM, Josh Boyer wrote:
On Fri, Jul 10, 2015 at 11:26 AM, Joe Brockmeier jzb@redhat.com wrote:
On 07/10/2015 12:59 PM, Josh Boyer wrote:
The atomic image is squarely targeted at being small, and for running containers. It is somewhat positioned as a CoreOS solution. With that being the case, I'm curious how the cloud image is different and not a repetitive image simply not using the atomic mechanisms.
That's sort of a key difference -- atomic == I can't just dnf install things. cloud == I can add on what I want the way I'm used to doing. (e.g., not containerized)
Yes, absolutely. However, if that is the only difference then I'm not sure how compelling it is when you compare it to all the other images provided elsewhere that let you do that already. Conversely, atomic is compelling _because_ of the Atomic platform. Atomic has novelty (for now), decent technical advantages, and a lot more marketing behind it.
Well, I mean... it's compelling for us because we want people to have Fedora available $all_the_places, right?
Not always. We don't want people to have Fedora on their phones. :)
We don't? I mean, I do - that doesn't mean we're invested in that, specifically, but that'd be cool. I'd certainly buy a Fedora phone. :-)
So having only Atomic means we'd basically be saying if you want to do things in the cloud, either do them the "Atomic way" or use another project, right?
The perception I've seen already indicates we're going that way. If all the hype is around containers these days, even the Fedora cloud download page plays Atomic up. It positions Atomic as the solution for containers and makes no mention of the fact that the base image would work too. It just says it's flexible.
It is. There are a number of scenarios where I can't imagine someone adopting Atomic right now. Remember, we don't even know how many people may be consuming the cloud image quietly...
I think we'd be sending the wrong message by abandoning the generic cloud image, though.
Sure, maybe. So instead maybe try sending just as strong of a message for the Cloud image as is done for the Atomic image. This is really primarily a marketing issue. The more I try and figure this all out, the more I think Cloud is being overshadowed by Atomic and keeping them together is detrimental in the long run.
My perspective is going to be much different from someone that lives and breaths cloud on a daily basis. But if I'm confused, there are other people out there wondering the same thing and clearing it up will help more than just me.
So, I guess I'm unclear on the ask here? What is the desired outcome you're looking for? Separate workgroups? A new product specifically around Atomic, or...?
On Fri, Jul 10, 2015 at 12:08 PM, Joe Brockmeier jzb@redhat.com wrote:
On 07/10/2015 05:01 PM, Josh Boyer wrote:
On Fri, Jul 10, 2015 at 11:51 AM, Joe Brockmeier jzb@redhat.com wrote:
On 07/10/2015 04:42 PM, Josh Boyer wrote:
On Fri, Jul 10, 2015 at 11:26 AM, Joe Brockmeier jzb@redhat.com wrote:
On 07/10/2015 12:59 PM, Josh Boyer wrote:
The atomic image is squarely targeted at being small, and for running containers. It is somewhat positioned as a CoreOS solution. With that being the case, I'm curious how the cloud image is different and not a repetitive image simply not using the atomic mechanisms.
That's sort of a key difference -- atomic == I can't just dnf install things. cloud == I can add on what I want the way I'm used to doing. (e.g., not containerized)
Yes, absolutely. However, if that is the only difference then I'm not sure how compelling it is when you compare it to all the other images provided elsewhere that let you do that already. Conversely, atomic is compelling _because_ of the Atomic platform. Atomic has novelty (for now), decent technical advantages, and a lot more marketing behind it.
Well, I mean... it's compelling for us because we want people to have Fedora available $all_the_places, right?
Not always. We don't want people to have Fedora on their phones. :)
We don't? I mean, I do - that doesn't mean we're invested in that, specifically, but that'd be cool. I'd certainly buy a Fedora phone. :-)
You would be disappointed.
So having only Atomic means we'd basically be saying if you want to do things in the cloud, either do them the "Atomic way" or use another project, right?
The perception I've seen already indicates we're going that way. If all the hype is around containers these days, even the Fedora cloud download page plays Atomic up. It positions Atomic as the solution for containers and makes no mention of the fact that the base image would work too. It just says it's flexible.
It is. There are a number of scenarios where I can't imagine someone adopting Atomic right now. Remember, we don't even know how many people may be consuming the cloud image quietly...
Awesome. Highlight some of those scenarios? Right now, the download page looks like this:
Cloud: We make this. Atomic: USE THIS FOR CONTAINERS
Atomic has a clear, highlighted usecase, to the degree that it is positioned as _the_ solution to that problem even though the base Cloud image would work just fine as well.
I think we'd be sending the wrong message by abandoning the generic cloud image, though.
Sure, maybe. So instead maybe try sending just as strong of a message for the Cloud image as is done for the Atomic image. This is really primarily a marketing issue. The more I try and figure this all out, the more I think Cloud is being overshadowed by Atomic and keeping them together is detrimental in the long run.
My perspective is going to be much different from someone that lives and breaths cloud on a daily basis. But if I'm confused, there are other people out there wondering the same thing and clearing it up will help more than just me.
So, I guess I'm unclear on the ask here? What is the desired outcome you're looking for? Separate workgroups? A new product specifically around Atomic, or...?
Short term, better marketing, so that the Cloud image isn't just "we make this. It's flexible and junk and stuff." Some description of use cases would be good. I mean, that is the question I asked at the start of the thread. The Cloud WG _knows_ all of this and I get the distinct impression that the expectation is everyone else in the world knows it too. Help the people like me that are perhaps just getting their feet wet know that Cloud is something worth trying.
Longer term, a new product around Atomic would make a lot of sense to me, yes. Particularly as the technology grows outside the scope of cloud anyway (see the desktop list yesterday for a discussion around atomic based desktop).
josh
On Fri, Jul 10, 2015 at 04:51:15PM +0100, Joe Brockmeier wrote:
provided elsewhere that let you do that already. Conversely, atomic is compelling _because_ of the Atomic platform. Atomic has novelty (for now), decent technical advantages, and a lot more marketing behind it.
Well, I mean... it's compelling for us because we want people to have Fedora available $all_the_places, right? So having only Atomic means we'd basically be saying if you want to do things in the cloud, either do them the "Atomic way" or use another project, right?
So, another option here is to twist an early decision we made by 90°. Switch from Cloud, Server, Workstation to Atomic, Server, and Workstation (sorry, branding/design team!). Atomic can run on bare metal, and we could start making Fedora Server cloud image and push those to Amazon, etc., and this could serve as the generic building-block cloud image people want. It'd probably be bigger, but that's okay because we'd know _why_. (And I hope with the improvements we're making on size in general, it wouldn't be too much bigger.)
So what I'm really after is what sets Fedora Cloud apart from every other distro cloud image. What usecases is it better at than {Ubuntu, SuSE, <whatever>}.
Given that logic, Fedora should stop everything but Atomic. The Cloud image should be Fedora optimized for the cloud instance experience, just like Workstation is Fedora optimized for the desktop user experience. It shouldn't be massively different for Cloud than Server, b/c the use case between Server and Cloud isn't that large.
Fedora should have a "typical" answer for what use cases are better than XYZ distro, that isn't dependent on a (frankly) edge use case like a container specific platform. Atomic is a new and interesting thing, with a very small and specific purpose and design. That's a good thing and shouldn't be used as an argument against the Cloud image. Or even as a comparison.
Depends: for end-users, it could mean a smaller bill each month on
storage.
I'm not a fan of this argument for minimizing the Cloud experience as the real cost of "magnetic" storage in most cloud providers is small. If pulling Python saves 1GB of on disk installed OS space, then users in a AWS environment save $0.24 a month / server in the most expensive storage in the most expensive region (Sao Paolo if you're curious). And I'm sure we aren't shaving that much off the image. I have to think the level of engineering required to majorly redesign things around minimization efforts are likely mis-placed if end user cost is the main metric.
That's where the Stack&Env WG work is important for us, as it could become
an asset against our other offers.
This ^^^ I'm a firm believer that the SCL work that got dropped is a huge value when we want to talk about differentiating Fedora as a Cloud or Server platform. The ability to cleanly separate system requirements from end-user platforms is huge. I think the Cloud SIG should be jumping up and down on getting SCLs back on track.
- Matt M
On Fri, Jul 10, 2015 at 11:42 AM, Josh Boyer jwboyer@fedoraproject.org wrote:
On Fri, Jul 10, 2015 at 11:26 AM, Joe Brockmeier jzb@redhat.com wrote:
On 07/10/2015 12:59 PM, Josh Boyer wrote:
The atomic image is squarely targeted at being small, and for running containers. It is somewhat positioned as a CoreOS solution. With that being the case, I'm curious how the cloud image is different and not a repetitive image simply not using the atomic mechanisms.
That's sort of a key difference -- atomic == I can't just dnf install things. cloud == I can add on what I want the way I'm used to doing. (e.g., not containerized)
Yes, absolutely. However, if that is the only difference then I'm not sure how compelling it is when you compare it to all the other images provided elsewhere that let you do that already. Conversely, atomic is compelling _because_ of the Atomic platform. Atomic has novelty (for now), decent technical advantages, and a lot more marketing behind it.
So what I'm really after is what sets Fedora Cloud apart from every other distro cloud image. What usecases is it better at than {Ubuntu, SuSE, <whatever>}. How should it be positioned so the people want to use it over those, etc. Fedora, in the Cloud space, is still behind in market share compared to the rest of the Linux world. I'm really curious if that can be overcome, or if instead the focus should be on Atomic entirely because it has a better chance.
(To be fair, Workstation and Server also have the same questions to answer in comparison to their peers. However, they have both history and familiarity on their side. Fedora has traditionally been a "desktop" OS and Server can feed off of RHEL which dominates the enterprise space. Cloud doesn't share that luxury.)
josh _______________________________________________ cloud mailing list cloud@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/cloud Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct
On 07/10/2015 04:59 PM, Matt Micene wrote:
I think the Cloud SIG should be jumping up and down on getting SCLs back on track.
What would this entail? I haven't been following that closely. I'm aware that SCLs have been... contentious.
TBH, I'm not sure what can be done from the outside either.
Not part of either the SCL team or FPC, all I see is what tickets and meeting logs I can find. So I'm not close enough to the issues either side is bringing up to understand if there's a "customer" argument to be made for "perfect is the enemy of good" in this case.
I just know I really like SCLs, how they work, and the capabilities. It's a big differentiator for the distro at large, and could be really useful for the envisioned Cloud end-user. I don't want to step on nuanced discussions of over 18+ months though.
- Matt M
On Fri, Jul 10, 2015 at 12:04 PM, Joe Brockmeier jzb@redhat.com wrote:
On 07/10/2015 04:59 PM, Matt Micene wrote:
I think the Cloud SIG should be jumping up and down on getting SCLs back on track.
What would this entail? I haven't been following that closely. I'm aware that SCLs have been... contentious. -- Joe Brockmeier | Community Team, OSAS jzb@redhat.com | http://community.redhat.com/ Twitter: @jzb | http://dissociatedpress.net/
cloud mailing list cloud@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/cloud Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct
On 07/10/2015 12:39 PM, Matt Micene wrote:
I just know I really like SCLs, how they work, and the capabilities. It's a big differentiator for the distro at large, and could be really useful for the envisioned Cloud end-user. I don't want to step on nuanced discussions of over 18+ months though.
If I'm not mistaken, though - the 'nuanced discussions' are pretty much at a standstill at this point?
Best,
jzb
It appears that way. No real new info on tickets I can find. Could be some discussion in lists (Envs?) or IRC.
- Matt M
On Mon, Jul 13, 2015 at 9:18 AM, Joe Brockmeier jzb@redhat.com wrote:
On 07/10/2015 12:39 PM, Matt Micene wrote:
I just know I really like SCLs, how they work, and the capabilities. It's a big differentiator for the distro at large, and could be really useful for the envisioned Cloud end-user. I don't want to step on nuanced discussions of over 18+ months though.
If I'm not mistaken, though - the 'nuanced discussions' are pretty much at a standstill at this point?
Best,
jzb
Joe Brockmeier | Community Team, OSAS jzb@redhat.com | http://community.redhat.com/ Twitter: @jzb | http://dissociatedpress.net/
2015-07-10 17:59 GMT+02:00 Matt Micene nzwulfin@gmail.com:
So what I'm really after is what sets Fedora Cloud apart from every
other distro cloud image. What usecases is it better at than {Ubuntu, SuSE, <whatever>}.
Given that logic, Fedora should stop everything but Atomic. The Cloud image should be Fedora optimized for the cloud instance experience, just like Workstation is Fedora optimized for the desktop user experience. It shouldn't be massively different for Cloud than Server, b/c the use case between Server and Cloud isn't that large.
Fedora should have a "typical" answer for what use cases are better than XYZ distro, that isn't dependent on a (frankly) edge use case like a container specific platform. Atomic is a new and interesting thing, with a very small and specific purpose and design. That's a good thing and shouldn't be used as an argument against the Cloud image. Or even as a comparison.
Depends: for end-users, it could mean a smaller bill each month on
storage.
I'm not a fan of this argument for minimizing the Cloud experience as the real cost of "magnetic" storage in most cloud providers is small. If pulling Python saves 1GB of on disk installed OS space, then users in a AWS environment save $0.24 a month / server in the most expensive storage in the most expensive region (Sao Paolo if you're curious). And I'm sure we aren't shaving that much off the image. I have to think the level of engineering required to majorly redesign things around minimization efforts are likely mis-placed if end user cost is the main metric.
Multiply this by the number of instances, and it could amount to few thousands dollars per month. More globally, our cloud image is still too fat compared to other distros by a large margin.
That's where the Stack&Env WG work is important for us, as it could become
an asset against our other offers.
This ^^^ I'm a firm believer that the SCL work that got dropped is a huge value when we want to talk about differentiating Fedora as a Cloud or Server platform. The ability to cleanly separate system requirements from end-user platforms is huge. I think the Cloud SIG should be jumping up and down on getting SCLs back on track.
- Matt M
Well, SCL is another topic (to remain polite), and I'm more than happy to let Stack&Env WG dealing with that matter.
Regards, H.
Multiply this by the number of instances, and it could amount to few thousands dollars per month.
Agreed, that could be an issue.
More globally, our cloud image is still too fat compared to other distros by a large margin.
I'd like to see data. The end user will wind up paying for all storage attached to a running instance, not just what OS is laid down. Based on a quick launch, the current Ubuntu 14.04 HVM community image in EC2 is based on an 8GB image.
What's a data supported target for an appropriate "small enough" sized Cloud image?
On Fri, Jul 10, 2015 at 12:04 PM, Haïkel hguemar@fedoraproject.org wrote:
2015-07-10 17:59 GMT+02:00 Matt Micene nzwulfin@gmail.com:
So what I'm really after is what sets Fedora Cloud apart from every
other distro cloud image. What usecases is it better at than {Ubuntu, SuSE, <whatever>}.
Given that logic, Fedora should stop everything but Atomic. The Cloud image should be Fedora optimized for the cloud instance experience, just like Workstation is Fedora optimized for the desktop user experience. It shouldn't be massively different for Cloud than Server, b/c the use case between Server and Cloud isn't that large.
Fedora should have a "typical" answer for what use cases are better than XYZ distro, that isn't dependent on a (frankly) edge use case like a container specific platform. Atomic is a new and interesting thing, with a very small and specific purpose and design. That's a good thing and shouldn't be used as an argument against the Cloud image. Or even as a comparison.
Depends: for end-users, it could mean a smaller bill each month on
storage.
I'm not a fan of this argument for minimizing the Cloud experience as the real cost of "magnetic" storage in most cloud providers is small. If pulling Python saves 1GB of on disk installed OS space, then users in a AWS environment save $0.24 a month / server in the most expensive storage in the most expensive region (Sao Paolo if you're curious). And I'm sure we aren't shaving that much off the image. I have to think the level of engineering required to majorly redesign things around minimization efforts are likely mis-placed if end user cost is the main metric.
Multiply this by the number of instances, and it could amount to few thousands dollars per month. More globally, our cloud image is still too fat compared to other distros by a large margin.
That's where the Stack&Env WG work is important for us, as it could become
an asset against our other offers.
This ^^^ I'm a firm believer that the SCL work that got dropped is a huge value when we want to talk about differentiating Fedora as a Cloud or Server platform. The ability to cleanly separate system requirements from end-user platforms is huge. I think the Cloud SIG should be jumping up and down on getting SCLs back on track.
- Matt M
Well, SCL is another topic (to remain polite), and I'm more than happy to let Stack&Env WG dealing with that matter.
Regards, H.
cloud mailing list cloud@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/cloud Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct
More globally, our cloud image is still too fat compared to other distros by a large margin. What's a data supported target for an appropriate "small enough" sized Cloud image?
OK, so I got curious. This is based on AWS, US West (Oregon). I compared current F22, Ubunutu 14.04, and openSuse 13.2. Here's the details:
Fedora 22 Base: Default image size: 6 GB [fedora@ip-10-0-0-176 ~]$ sudo fdisk -l Disk /dev/xvda: 6 GiB, 6442450944 bytes, 12582912 sectors
Root filesystem: 444M used /dev/xvda1 ext4 5.9G 444M 5.2G 8% /
Ubuntu 14.04: Default image size: 8 GB ubuntu@ip-10-0-0-68:~$ sudo fdisk -l Disk /dev/xvda: 8589 MB, 8589934592 bytes
Root filesystem: 780M used /dev/xvda1 ext4 7.8G 780M 6.6G 11% /
openSuse 13.2: Default image size: 10 GB ip-10-0-0-175:~ # fdisk -l Disk /dev/hda: 10 GiB, 10737418240 bytes, 20971520 sectors
Root filesystem: 868M used /dev/hda1 ext4 8.8G 868M 7.5G 11% /
(AND openSuse make you log in as root!! eep!)
So we are currently the smallest uploaded AMI and consuming the least amount of available space in base, current version, community AMIs.
On Fri, Jul 10, 2015 at 12:46 PM, Matt Micene nzwulfin@gmail.com wrote:
Multiply this by the number of instances, and it could amount to few
thousands dollars per month.
Agreed, that could be an issue.
More globally, our cloud image is still too fat compared to other distros by a large margin.
I'd like to see data. The end user will wind up paying for all storage attached to a running instance, not just what OS is laid down. Based on a quick launch, the current Ubuntu 14.04 HVM community image in EC2 is based on an 8GB image.
What's a data supported target for an appropriate "small enough" sized Cloud image?
On Fri, Jul 10, 2015 at 12:04 PM, Haïkel hguemar@fedoraproject.org wrote:
2015-07-10 17:59 GMT+02:00 Matt Micene nzwulfin@gmail.com:
So what I'm really after is what sets Fedora Cloud apart from every
other distro cloud image. What usecases is it better at than {Ubuntu, SuSE, <whatever>}.
Given that logic, Fedora should stop everything but Atomic. The Cloud image should be Fedora optimized for the cloud instance experience, just like Workstation is Fedora optimized for the desktop user experience. It shouldn't be massively different for Cloud than Server, b/c the use case between Server and Cloud isn't that large.
Fedora should have a "typical" answer for what use cases are better than XYZ distro, that isn't dependent on a (frankly) edge use case like a container specific platform. Atomic is a new and interesting thing, with a very small and specific purpose and design. That's a good thing and shouldn't be used as an argument against the Cloud image. Or even as a comparison.
Depends: for end-users, it could mean a smaller bill each month on
storage.
I'm not a fan of this argument for minimizing the Cloud experience as the real cost of "magnetic" storage in most cloud providers is small. If pulling Python saves 1GB of on disk installed OS space, then users in a AWS environment save $0.24 a month / server in the most expensive storage in the most expensive region (Sao Paolo if you're curious). And I'm sure we aren't shaving that much off the image. I have to think the level of engineering required to majorly redesign things around minimization efforts are likely mis-placed if end user cost is the main metric.
Multiply this by the number of instances, and it could amount to few thousands dollars per month. More globally, our cloud image is still too fat compared to other distros by a large margin.
That's where the Stack&Env WG work is important for us, as it could
become an asset against our other offers.
This ^^^ I'm a firm believer that the SCL work that got dropped is a huge value when we want to talk about differentiating Fedora as a Cloud or Server platform. The ability to cleanly separate system requirements from end-user platforms is huge. I think the Cloud SIG should be jumping up and down on getting SCLs back on track.
- Matt M
Well, SCL is another topic (to remain polite), and I'm more than happy to let Stack&Env WG dealing with that matter.
Regards, H.
cloud mailing list cloud@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/cloud Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct
On Fri, Jul 10, 2015 at 02:30:43PM -0400, Matt Micene wrote:
So we are currently the smallest uploaded AMI and consuming the least amount of available space in base, current version, community AMIs.
Now there's a quote. :)
Now we can focus on making the Docker Base image as small as possible to include heroic efforts ;-)
On Fri, Jul 10, 2015 at 3:07 PM, Haïkel hguemar@fedoraproject.org wrote:
Awesome, it wasn't the case a year ago.
H.
cloud mailing list cloud@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/cloud Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct
-
Haïkel -
Joe Brockmeier -
Josh Boyer -
Major Hayden -
Matt Micene -
Matthew Miller -
Ryan Brown