Your Notion Of Cloud Hosting?

HostColor

Member
Hello,

Hope you are having nice holidays and have time to talk thumbup:

Let me wish you to play part in a fast growing web hosting industry in 2010 and of course to take advantage of this.

Now on the topic. How do you understand "Cloud Hosting"? What kind of infrastructure and platform do you imagine when someone mentions Cloud hosting?

I would suggest 6 "fields" to be filled with answers:

Operating system:
Virtualization: Yes; No (What kind?)
Software:
Network:
Data center:
Other features:

I have my answer, but I'll share it with you later!
 
You should include Instances in your list also. Remember a TRUE REDUNDANT cloud will be in multiple data centers.

For me, I see too many hosting companies attempting to run their own cloud, or offer cloud hosting, and operate out of a single data center facility. YES it likely does satisfy the requirements to be "cloud" but really, the purpose is to have instances in various parts of the world to serve the data faster. it certainly costs more to do this, but when you're dealing with enterprise sites, you get what you pay for.

I don't pretend to know all about the cloud - it's too new and seems to be more "concepts" to many places than anything else. So I'll be watching the thread to see what other insights I can glean. I've read many (too many) articles and whitesheets on Cloud Hosting and no sooner do you put it down, than another is released contradicting it :)
 
The cloud sounds great in theory but to implement proper cloud hosting infrastructure is very expensive
 
Hi Conor,

I shall disagree that a "TRUE REDUNDANT cloud will be in multiple data centers". this is only an option. Operating equipment in one infrastructure in different data centers is CDN, something businesses did year before the concept of Cloud computing to appear.

If you use global redundant network for Cloud Hosting service, you do not need to have infrastructure in different physical locations, unless you really need some kind of localization similar to Google local search. If you are service provider, you do not need this.
 
Can you define "global redundant network"?

If you do not need to have data in different locations, if a data center goes offline (like they do - it's not UNCOMMON), how does the data stay active for viewers on the web? Doesn't the data need to be replicated to an outside machine SOMEWHERE?
 
a notion on cloud hosting

I'll throw my hat in the ring here.

I've been in the hosting business for over 6 years. Needless to say, I'd compare cloud hosting with the shared hosting methodology of the late 1990's to early 2000's. Today however, Cloud Computing has become a much different animal. There are higher levels of security, performance and manageability that are defining what cloud computing truly should be and is becoming right now.

Anyway, to your questions;

1. OS: I really think the OS selection is based on the capabilities of the provider and their ability to support those needs with experts. In my mind, Cloud Computing should offer both Microsoft and Linux based operating systems.

2. Virtualization: This is a piece of the puzzle. Right now, VMWARE, Citrix and Parallels are the only companies providing what I'd say is an easy to deploy platform to offer a scalable and secure computing platform. In the future, the underlying virtualization technology will matter less when API's and customization become more prevalent. At this moment, I'd say that Citrix and VMWare will dominate for quite some time because of their financial capabilities and their general acceptance as reliable products. Although Microsoft and Google will have something to say about that.

3. Software: Id say any development platform should be built to live in a multi-tenant configuration and can easily scale across multiple processors.

4. Network: This is a big thing and the cloud most certainly should have more than 1 Tier-1 (Verizon,ATT, Level3) provider connected to it. As someone mentioned earlier, geo-diversity- or federated cloud- will build a truly resilient network for maximum uptime. Look for this from hosting.com in 2010.

5. Datacenter: Tier3 or better data center. Multiple carrier access, N+1 or better power and cooling. 24x7x365 support.

6. Other features: Well, API support, geographic load balancing, easy to use customer interface (Self Service).
 
Can you define "global redundant network"?

If you do not need to have data in different locations, if a data center goes offline (like they do - it's not UNCOMMON), how does the data stay active for viewers on the web? Doesn't the data need to be replicated to an outside machine SOMEWHERE?

I'll give an example. Having a good and stable connections with 2 or 3 major U.S. carriers + NTT and another one to Asia and 2 more to Europe... (let's say Tiscali and Interoute) will be enough to say you have a "global redundant network".
 
So "global redundant network" is not the same as a "global redundant site" then. You're just looking at multiple carriers for the data. If the data center goes offline (network issue, power issue, someone trips over the power cord (ahem - rackspace), or the electric room catches on fire (ahem - the planet), or the basement is flooded (uhh.. can't remember the datacenter, but it was in Chicago) - so those items don't necessarily play into the roll of a redundant NETWORK - these relate to the SITE in particular.

See, for me, my idea of redundant would be a multi-location site where if your site goes offline at one place, you're still online. This is what has been broadcasted on a number of places offering CLOUD and how stable and superior Cloud really is. Where in fact it's nothing more than shared hosting with the ability to increase processing power, disk space, memory etc all on the fly.

Let me know if I got that right.
 
The concept of cloud computing isn’t new considering the concept goes back as far as the 60’s though the way in which it is being used by providers is in my opinion. The whole point of cloud computing is from a hosting standpoint to provide speed, stability and redundancy across as wide an area as possible in order to increase the benefit for the potential client base, done by virtualization.

However brining in the point that Conor made, the whole point of a cloud network is to provide a redundant network across multiple locations in order to avoid many of the problems of traditional systems and combine them with the benefits of the similar VPS technologies.Otherwise the effectiveness of the cloud within a single datacentre is simply to provide an expandable VPS solution mirrored across multiple machines as essentially it would have similar redundancy for many of the issues that cause us as providers downtime.

The use of multiple transit providers does little to nothing to provide redundancy if a primary switch on the network has a malfunction for example or any of the examples provided above, as such the virtualisation layer of the cloud network ensures that the data is mirrored across multiple sites and an alternate site would take or share the load with other sites in order for the users site to remain available and unaffected by the malfunction or natural disaster.

Where my knowledge is lacking is the information from scripts that are held in RAM or being processed which could lead to corruption, though there are a number of solutions I’m yet to read that in-depth to any particular approach :)
 
Sure,

There's only one thing that I would like to point out and it is that having infrastructure and redundant network across multiple locations IS NOT part of the "Cloud" concept. However I shall admit that if a company operates 2 or more facilities in a CDN, which is part of a cloud platform and/or service is something that shall be appreciated from its customers.
 
There are dozens of other examples of hosting providers doing cool things with the cloud (an exhaustive review would be too much of a stretch), and many other somewhat dubious examples of companies putting the cloud badge on whatever they could. Certainly, there is still a discussion taking place over what exactly ought to be considered "cloud computing."
Thought this was very interesting - from an article in The WHIR
 
I think that the Could won't have real application to end users that only require their web sites online. I think that the key here is that the cloud will be of great utility for developers that require intensive access to their applications and are in need of platforms that can support peak usages and add resources the time they need, not paying for month or what they don't need.
 
You should include Instances in your list also. Remember a TRUE REDUNDANT cloud will be in multiple data centers.

For me, I see too many hosting companies attempting to run their own cloud, or offer cloud hosting, and operate out of a single data center facility. YES it likely does satisfy the requirements to be "cloud" but really, the purpose is to have instances in various parts of the world to serve the data faster. it certainly costs more to do this, but when you're dealing with enterprise sites, you get what you pay for.

I don't pretend to know all about the cloud - it's too new and seems to be more "concepts" to many places than anything else. So I'll be watching the thread to see what other insights I can glean. I've read many (too many) articles and whitesheets on Cloud Hosting and no sooner do you put it down, than another is released contradicting it :)

Yes correct, you can provide redundancy with VMware SRM
 
Back
Top