It has been very interesting reading the various comments is response to Dominic's blog post, but I have to say I have to agree with @neilcattermull I am also seeing the exact opposite to the point put across by Dominic. I think we can all agree that one of the main drivers behind virtualisation was the fact that physical servers were highly inefficient and their computing resources massively underutilised. If you want any kind of redundancy in the physical world you have to by two of everything, which doubles your cost straight away. The cloud offers small and medium sized businesses the ability to access enterprise class IT systems at a price they can afford. In my experience businesses today do not want to pay for anything they don't have to, that is why the cloud 'pay as you go' model is so attractive. Traditionally high CAPEX prevented many organisations from investing into new infrastructure and upgrading systems etc… Now with the emergence of Cloud Service Providers like ourselves those previously prohibitive costs can be translated into monthly OPEX costs. At the end of the day virtual or physical servers are just one piece of the puzzle for the customer. There end goal is to ensure that their applications run smoothly and their able to expand their infrastructure as their business demands. So the question remains, is the cloud better equipped to deliver on this, in my opinion I would say yes.

Don’t get me wrong, cloud isn’t completely dead! There are numerous benefits to the cloud that we continue to sell to our customers. At the same time though, for certain customers, physical hosting infrastructure makes more sense. Cloud can be bloody expensive both for us and our customers which makes a physical solution more appealing in certain cases and more cost effective. And as for security, sorry @AdamMoss but that isn’t a concern for us. Our datacentres are extremely secure physically meaning all of our customers can sleep well at night. The data within our datacentres isn’t moving anywhere regardless whether a customer uses cloud or dedicated hosting.

Apologies for the comment system failing our third party provider Livefyre had some server issues. This situation we believe is now rectified.

My latest conversation: D-Link® to Expand its Channel Solutions with SaaSMAX Platform

Some great perspective from Dominic Monkhouse as well as in the comments that followed; particularly those from meizlik when he speaks about costs. Most comments about cost have focused only on the infrastructure and equipment costs. How about the savings in support costs that are realized as well as a lot of that overall risk that is transfered in the SLA over to the IaaS provider? We provision and host data vaults at 6fusion iNodes all over NA. As a smaller company, in addition to realizing the typical benefits of scalability and agility, global reach, If they really mess up, there are penalties and we can take legal recourse if necessary. If we staffed this ourselves what's the most I could do, fire someone? IMHO risk mitigation and transfer are strong but often overlooked selling points of cloud hosting as long as proper due diligence is completed.

Cloud isn't right for everyone in every scenario and no one cloud/virtualisation outcome fits all either. So we can expect to see customers move between cloud formats as they find their feet over time for different application needs. The trend is still to see on network moving to cloud and few go back with surveys showing a high customer satisfaction from those adopting cloud (eg Cloud Industry Forum survey showed a high 9x% CSAT). Ian Moyse Workbooks.com

Great points made across the board. In a nutshell, I think the reason everyone is saying "it depends" is because not all costs have the same value to different people. To a larger company with predictable loads, the cloud may prove more expensive as my friend Justin points out. Btw, Hi Justin - hope you're doing well! However, to a small startup that lacks a predictable load, the cloud may prove less expensive. But even more important is that the value of having an instantly scalable cloud may actually be worth more to the startup than the dollar savings from any infrastructure investment. You see, how the infrastructure supports your business may actually prove more valuable than the infrastructure costs, and I think it's that comparison that each business needs to make on its own and is a great basis for Dominic's argument. Ditlev illustrates this when he says the cloud may be a "great way of dealing with redundancy." If that's of value to a company, it may change the numbers considerably. Dave Meizlik with Dome9 Security www.dome9.com

In my view the actual answer is very much that "it depends" as not every workload suits being moved to a third party provider whether it's outsourced, hosted or on a shared platform - there is no reason why you cannot move to a flexible "hybrid" service delivery model. Many customers are gradually making the transition by moving standard commoditised services like file, email or SharePoint to external cloud services whilst retaining other more complex or critical services in-house - note running services in-house doesn't mean you can't procure them as a service, its simply that the location is different. The ability to ramp up services instantaneously is also an interesting concept and has tremendous value for the type of organisations that require that level of rapid response. However many companies value a solid security model, proven change control and guaranteed service availability over the ability to switch a thousand VM's on and off to meet changing demand. One size doesn't fit all and it's worth mentioning that there is more to life than cost. Many organisations move to a Service Provider/Cloud/Outsource model to mitigate risk or simply because they gain flexibility and it frees their staff to focus on product development and innovation rather than managing spinning disks, backup tapes and VM's.

From Justin Pirie Director of Communities and Content www.mimecast.com There are definitely some edge cases like Ditlev outlines, where with predictable demand, it's much cheaper to use dedicated tin, like we do here at Mimecast. Our loads are relatively constant and as we add customers we add boxes to service them. It's hard to know from the technical details that were given above- what type of "cloud" they were using previously, as so much is Cloudwashed these days. I'd also have thought that an online gaming company would not want physical tin if their game was a breakout success. In regards to private cloud- Zynga got to a scale where using Amazon was no longer cost effective- so they built their own private cloud to replace it. http://wikibon.org/wiki/v/Zynga_Cloud_Case_Study:_The_journey_to_a_real_private_cloud However, I'm seeing more of what Neil Cattermull is, more and more people demanding on-demand infrastructure, even if what they're really just after is virtualisation. What you can't discount however is people using older techniques to solve current problems, with physical infrastructure and you cannot blame Dominic for solving their business needs and taking their money- that's his job! Those architects who put together that design might not survive long at Netflix/Zynga/Instagram etc...

There are definitely some edge cases like Ditlev outlines, where with predictable demand, it's much cheaper to use dedicated tin, like we do here at Mimecast. Our loads are relatively constant and as we add customers we add boxes to service them. It's hard to know from the technical details that were given above- what type of "cloud" they were using previously, as so much is Cloudwashed these days. I'd also have thought that an online gaming company would not want physical tin if their game was a breakout success. In regards to private cloud- Zynga got to a scale where using Amazon was no longer cost effective- so they built their own private cloud to replace it. http://wikibon.org/wiki/v/Zynga_Cloud_Case_Study:_The_journey_to_a_real_private_cloud However, I'm seeing more of what Neil Cattermull is, more and more people demanding on-demand infrastructure, even if what they're really just after is virtualisation. What you can't discount however is people using older techniques to solve current problems, with physical infrastructure and you cannot blame Dominic for solving their business needs and taking their money- that's his job! Those architects who put together that design might not survive long at Netflix/Zynga/Instagram etc...

if - as a infrastructure consumer - your usage patterns are predictable, and without large peaks...then yes, a fully dedicated infrastructure will surely be better and cheaper than a virtualized cloudy setup. Public cloud is great if you have huge spikes that would leave most of your dedicated infrastructure un-utilized most of the time, and it's a great way of dealing with redundancy as you can share your failover resources with other users residing across the same servers. Though, that comes with a cost obviously - Service providers are expected to keep idle capacity sitting ready for clients to burst into or fail over to, and they will obviously not keep that capacity idle for free. So, the cost of the idle capacity will be shared across the userbase, but if you - as an end user - see no value in having the capacity available to you, then why should you opt-in and pay for it? The cloud business model turns everything up-side-down for service providers, they are used to oversell their infrastructure by xxx%, and now suddenly with the cloud service providers have to undersell and actually leave capacity available waiting for clients to buy it, without being able to charge for it. I sometimes compare that with the airline industry - the analogy is pretty clear. In shared hosting it's like fitting 1000+ passengers in a 500 seat jet, knowing that half of them will not turn up (ie. use the space allocated to them), in a dedicated server world you can be sure that every seat is paid for ... though in the cloud space you have to keep 20-40% of your seats unsold as your passengers may multiply midair: http://ditlev.onapp.com/PowerPoint_Slide_Show_-_%5Bonapp_karl.pptx%5D-20120807-141425.jpg Currently service providers charge a premium for that, and they can as the demand is certainly there, so they can add healthy margins to their cloudified infrastructure ... That's why Dom may actually have a point here. Down the line all that may change though, which is why it's key that service providers find alternative ways to monetize their spare capacity. So, as much as I love that picture Daniel, Dom really has a point! :) D

The future is "Pizza Hosting" same base, just different toppings. Cloud or dedicated it is now a major challenge sorting the differentiators between providers. I agree the trend is reverting back to dedicated, but only because the enduser is struggling to understand or educate internally the concept. The path of least resistance remains the norm...

400 visits and 12 comments we think Donald has got of lightly!

I would love to hear Dominic's technical and business explanation of how a return to this model would make economic sense.

Server hugging dinosaur might become the next in vogue pet!

Interesting blog guys. Some interesting points raised. Lovely picture too!!! Looking at it from a security angle. Outsourcing your infrastructure to a external supplier takes a large amount of trust. You have to be able to trust that the service provider is following the correct security procedures. I think there are many cloud customers realising that the "secure" cloud provider they signed up to is not as security conscious as they were originally lead to believe by the sales patter. Even if a cloud provider claims to be certified with a compliance model such as ISO 27001, this is not a guarantee that they are actually following the security procedures adequetly. As a result certain customers have decided to move back to in-house phsyical infrastructure set up and are sleeping more soundly at night knowing their servers and their data are locked away on their site and managed by their people. There are a number of cloud providers working to address these doubts and issues regarding security of cloud services and there are cloud providers out there who do provide real security but you have to do some background research on the providers before you dive in.

I would be interested to get a bit more on the cost-effectiveness of the various set ups: while I sort of get the argument over licensing for virtualisation I'm not seeing how using a part of shared IaaS can be less cost effective than using your own private infrastructure. Even on the licensing front, for anything with a spiky usage profile, it surely has to be more cost-effective in most cases to pay slightly more but only for the period of time that it is needed, which is the model that is seen as you move from tradionally licensed to SaaS.

BTW - Nice picture Dan :)

This is an interesting take from Dominic. On my travels I have seen the exact opposite to be honest. The client demands the flexibility of "pay as you go" technology with the ability to be flexible and able to adapt to business initiatives - Isnt this Cloud based patforms? I cannot see any benefit in static technology, even if said technology comes cheap. With the emergence of the multiple cloud SaaS, PaaS, IaaS providers, all I can see for the future is cloud technologies getting more and more established. With this said, hybrid cloud technologies seem to be the first journey to take up and I have not seen too many large cap firms make the biggest leap yet (except where they already outsource their technology to Cap Gem or similar organisation). I understand that some firms using cloud technologies/firms maybe unhappy with their current provider, however once you have migrated to a cloud provider/infrastructure, you rarley move back to static technology, just providers. In the future, all cloud service companies will be basically the same (a bit like cars now) but with different options and at the end of the day the main differentiator will be price. Nice blog Dan and I will be interested to see other comments

Feel free to comment using the Livefyre tool or just log in using twitter linked in or facebook!