There has been this article published recently on the Tech Republic. http://www.techrepublic.com/blog/10things/10-reasons-you-should-build-your-own-cloud/3637?tag=mantle_skin;content
It is ridiculous and misguided to the point of being dangerous (just at the point where real cases of savings are appearing, it suggests a backwards approach), but it has promoted some pretty interesting debate: I am simply copying some of the comments I found entertaining and insightful ;)
Can Google give the author his Google Apps on a CD? by: den123net What a moronic take on things! (Unless the article is a joke, in which case, its pretty good, as it caused solid laughs across a large office full of techies). I have now shared it across colleagues and even clients and all had a good belly chuckles..
There is no In-House Cloud by definition. The same way as there is no "National Power Grid" in-house. You can have a server room, sure, even with back-ups (wow), but it is not, nor ever be a Cloud
Will your own Cloud magically grow when you need more data / compute? Will it shrink at night when you don't need it?
The author should get a tech job, (writing does not count), and even then refrain from advising anyone on anything ...
This generated this response:
techies? by: rduncan@... Will your own Cloud magically grow when you need more data / compute? Will it shrink at night when you don't need it?
it won't grow magically, but neither does any cloud,however you can build a hybrid cloud and consume some public resources as you need them (is this magic?) - will it shrink at night - yes! this can be controlled by policy, not just virtual machines but right down to powering downing nodes, these are built in features of Intel Sandy Bridge & Ivy Bridge chipsets that private cloud developers are taking advantage of and can save even small data centers thousands per year.
Private cloud is the fastest growing industry on earth at the moment, makes me wonder about that large room of 'techies'
In turn, here is the reply:
Yes techies! As in people who used to build DCs, before cloud made DCs die
This debate is getting silly (it like arguing that horses are better than cars), but here is my last reply on this:
Again, there is no such thing as in-house or private Cloud. Like there is no such thing as an in-house power grid. You can host in-house and have your own private DC (on site or off site), but it is not, nor will ever be a Cloud. Calling it your cloud is silly.
it won't grow magically, but neither does any cloud, = yes it does . Well not magically, but companies like AWS, Google, MS have technology which took them years and billions to develop that you will not have access to. They can add capacity at such a rate (and they do) that the dont actually know how many servers they have (Google has stated this publicly) It does not matter to them. They have moved away from individual servers or racks or even DCs - it is all just capacity for them now. They run custom Hypervisors and OSes that you cannot deploy in house, and never will... So its like "magic" for most of us.
however you can build a hybrid cloud and consume some public resources as you need them (is this magic?) = Not it is not magic (see above), but, the economic argument is lost immediately if you accept this: If hybrid works, then why full Cloud does not? Demand is peaks and troughs. If you build DC for the max demand (the past) - at least 50% is wasted - when you dont use it. If you build it for average, you waste 25% only, but have to pay extra every time you go to max.
If you build for the minimum, then your utilisation is 100% (in theory), but then why make the investment at all?
Hopefully you get the point that the only possible rationale for hybrid cloud is when you have ALREADY (in the past) made the investment, and just want to migrate to Cloud gradually, and sweat something that you cannot do anything else with for a while. Even then it is dubious, which brings us to the next point.
will it shrink at night - yes! this can be controlled by policy, not just virtual machines but right down to powering downing nodes, these are built in features of Intel Sandy Bridge & Ivy Bridge chip sets that private cloud developers are taking advantage of and can save even small data centres thousands per year. = Lol. Really? You can turn down power and cooling, even turn off machines at night - but this is saving a fraction of the cost. Do you get some of you initial investment back? Do you shrink the room in the building? Do you shrink security / patches / maintenance / salaries of the employees managing it? No of course not. Thats why those costs are called "Fixed". In fact, even at peak times, every manual tells you not to run anything at 100% capacity - 75% is typically the top threshold. So what happens to the 25% remaining? Do you get that back? Does someone pay you? Rhetorical question...
Private cloud is the fastest growing industry on earth at the moment, makes me wonder about that large room of 'techies' = where did you get this from? Public Cloud (AWS, Google, MS, Rackspace and a few others) are growing exponentially, where as server and hardware sales are crashing (the stuff that would be used for "private clouds" (HP, IBM et al all whining and trying to build their own private clouds) - just do some simple googling or read anyone from Gartner to the The Register to see this.
The room of techies, most used to design DCs and build / deploy / manage servers, large and small. They have all left the industry to join the one that has come to replace it - Cloud.
I think the final one nails it, in that it covers the main argument point. In the end, it just progress. Every time we have a similar conversation with some of our clients, and they strongly disagree, I just relax - the Cloud revolution is happening - whether someone agrees with it or not... Like all progress, it is unstoppable, not matter how many wrong articles appear around the net...