This is an old revision of this page, as edited by LirazSiri (talk | contribs) at 03:49, 28 February 2010 (→Proposal: remove Servers). The present address (URL) is a permanent link to this revision, which may differ significantly from the current revision.
Revision as of 03:49, 28 February 2010 by LirazSiri (talk | contribs) (→Proposal: remove Servers)(diff) ← Previous revision | Latest revision (diff) | Newer revision → (diff)| This is the talk page for discussing improvements to the Cloud computing template. |
|
Archives: 1, 2Auto-archiving period: 30 days  |
| This is the talk page for discussing improvements to the Cloud computing template. |
|
| Archives: 1, 2Auto-archiving period: 30 days |
Template Purpose
This template was created for use on the cloud computing article as an illustration by example of the top level taxonomy or "layers" of same. It was not intended to advertise specific products nor be extended with arbitrary categories, rather to give a small number (5-10) of the best examples available. Documentation updated accordingly. -- samj in 16:43, 24 February 2010 (UTC)
Relevancy
I am not sure Facebook or Android have to do with cloud computing... IMHO cleanup required.--Kozuch (talk) 10:15, 14 August 2008 (UTC)
- Agree with Facebook, the apps aren't run in the cloud. Sure, they're internet apps but they're run by the developers. --TheSeer (TalkˑContribs) 06:51, 14 April 2009 (UTC)
- Ok, works for me. -- samj in 20:45, 22 February 2010 (UTC)
What goes in?
I've just purged a couple of non-notable entries in the template. Anything that goes into the template pops up across many pages, so we should be much stricter about what goes in here than anything else.
- Things should be discussed here before being added, especially whole new groups
- Only highly notable examples should go in.
- Stuff that goes in without discussion ought to be pulled and discussed, especially if there is any sign of CoI issues.
Thoughts? SteveLoughran (talk) 13:41, 27 November 2009 (UTC)
- The point is to help people understand what each layer is/does, not to advertise solutions. Examples should be household names and there should only be a few (5-10) at each layer. For a start you can rip out anything without its own Misplaced Pages article, and even then check to see whether the article shouldn't be prodded or AfD'd. -- samj in 20:42, 22 February 2010 (UTC)
Proposals
These are missing the point - the idea was to illustrate each of the layers and provide a high level taxonomy, not a linkfarm/coatrack for any vendor that cares to associate themselves with cloud (which is pretty much every vendor by the way). -- samj in 20:45, 22 February 2010 (UTC)
Proposed Group: Algorithms
I propose a new group, algorithms. Distributed Hash Table, MapReduce, PageRank Column Table are all examples of datacentre-scale algorithms that merit a mention. By linking to the algorithms, we can stop people sticking in links to every implementation of DHT, MapReduce, etc. SteveLoughran (talk) 13:46, 27 November 2009 (UTC)
Agreed. LirazSiri (talk) 21:51, 29 November 2009 (UTC)
Proposed Group: Technologies
The other thing we could list is technology areas (As opposed to specific products)
Technologies would be areas of computing common to many views of what Cloud computing is : Virtualization, automated deployment, [[web services (in the purest form, not just WS-*), datacenters, distributed file systems. These aren't protocols, and they aren't specific products. Indeed, by listing the technologies we could ensure that implementations of specific technologies get added to the technology pages, rather than having people add links to their favourite/personal product in this template.
If nobody objects, I will start a listing of the above technologies. And explictly purge any attempts to add a single product into the listing. That is not a technology, it's a prodcut.
Sounds good. LirazSiri (talk) 21:49, 29 November 2009 (UTC)
Proposed Group: Appliances
Earlier I attempted to add an Appliances section which SteveLoughran removed with the following rational: "Templates may be part of the EC2 world, but not in other stacks.". Regardless of what you call them (or how they're implemented), pre-integrated software stacks are an important part of on-demand server provisioning. There's little point in provisioning a server that isn't configured for the role you need it. That's true whether you are provisioning a server in EC2, GoGrid, or on a private cloud (e.g., running VMWare infrastructure or Xen). Cloud computing users are just like everyone else in the IT world in that they want stuff to just work.LirazSiri (talk) 21:49, 29 November 2009 (UTC)
- I felt it was over biased towards EC2-as Cloud Computing to have a whole row of appliances in there. If we had a technology row on virtualisation it could be something that includes appliances as one entry, alongside PXE/Kickstart driven VM instantiation on demand from a list of RPM/deb files. SteveLoughran (talk) 09:13, 30 November 2009 (UTC)
- No no there's nothing EC2 specific about appliances. How you implement an appliance is not that important. What's important is that you're pre-integrating a system to align with a particular narrow usage scenario into some kind of package which can then reproduce the result on scale. How you distribute the result is a relatively minor practical detail. The result can be packaged as an ISO, a VMDK HD image, AMI, or a set of interpreted instructions such as a list of RPM packages and preseeded configuration variables. It's still an appliance. Cloud computing is mainly about economics, and cloud computing and appliances go hand in hand because the economics work out. Instead of integrating over and over again and having to test each inconsistent result separately you are leveraging economies of scale and put more resources into creating the very best integration of components and then "mass producing" it with minimum marginal cost. That's especially important when you are provisioning servers on the fly. LirazSiri (talk) 03:18, 2 December 2009 (UTC)
- I'm returning to this topic as I'm watching the conflict between Liraz and SamJ. I propose that we create the technologies section, create a "Cloud Appliances" page that can contain technology-neutral coverage of this area. Just listing various people who have tooling or appliance VMs misses the point, and will end up being a WP:EL troublespot on the category listing itself. Similarly, we need something for Cloud Infrastructure APIs that looks at the various APIs, lists some requirements (long haul, security), problems: politics, notifications, etc, and it can be somewhere the various APIs can be looked at. Let's resolve this conflict by giving VM appliances a name and a place, and index that under Cloud Technologies. SteveLoughran (talk) 23:23, 26 February 2010 (UTC)
- I agree that "technologies" is a good section to have but unless we further edit the cloud computing template substantially the template will be unbalanced. In other words there's an inconsistency with the way cloud appliances entries are handled vs other areas of the cloud computing stack such as "platforms" and "infrastructure" which do list specific projects and vendors. For example, why do specific infrastructure vendors such as FlexiScale and GoGrid or even an open source project such as Nimbus deserve as much weight as an entire group of appliance vendors? LirazSiri (talk) 03:37, 28 February 2010 (UTC)
- Wherever they go, they definitely don't belong here. -- samj in 23:40, 26 February 2010 (UTC)
Proposal: remove Servers
I have to agree with LirazShiri's actions at rm-ing the server line. We could have a datacenter hardware corner that looks at the hardware issues -and if you don't think there are problems in building power-efficient high availability datacentre hardware then you don't even know what the problems are, what PUE stands for or what limitations DNS encounters in this world. You may see things like AWS banning multicast IP but not understand why that is, or why vendor costs are the way they are. But having a row listing HW vendors is just doomed, everyone will end up battling over what goes in. SteveLoughran (talk) 00:11, 27 February 2010 (UTC)
- The source of the "Servers" and "Clients" layers is to capture the physical devices and operating systems that power and consume cloud computing services respectively. Without it you end up with things like CloudRack and Cisco UCS as well as virtualisation platforms like VMware trying to creep into the Infrastructure layer as well as mobiles, browsers, operating systems etc. creeping into Applications from above. Remember, this template should reflect/illustrate what is in the cloud computing article. -- samj in 02:25, 27 February 2010 (UTC)
- First, I disagree with the notion that the cloud computing template has to be some kind of table of contents for the cloud computing article. Rather than try to capture the one true taxonomy of cloud computing we should strive to maximize utility for our readers. In practice the template which is included on many cloud computing related articles serves as a resource for providing further links of interest to other related articles on Misplaced Pages. It's telling that the Server section doesn't even link to proper articles but rather to vendors. Also I stand by my original assertion that cloud computing has nothing to do with the hardware vendors listed. Cloud is a software/economic paradigm. You can implement "a cloud" with any hardware. Where do we draw the line? I bet the vendors listed only account for a tiny insignificant portion of the actual hardware used in a cloud computing configuration. Conversely you could use the hardware marketed for "Cloud computing" in a software/economic configuration that nobody would genuinely identify with cloud computing. The Servers category is a poor fit for this template. LirazSiri (talk) 03:49, 28 February 2010 (UTC)
- OK, but's not really "server" so much as "datacenter technologies and infrastructure". SteveLoughran (talk) 17:25, 27 February 2010 (UTC)