Cloud enablement

Clouds Need More Than SLAs

Pete Baldwin 12/5/2012 | 5:05:41 PM
re: Clouds Need More Than SLAs

So, this is a balancer to the theory Randy Bias has been espousing, about the enterprise cloud being a waste of time because Amazon-style commodity clouds can be had for so much cheaper.


And the two have discussed this with each other quite a lot.

Any thoughts? A lot of people (Heavy Reading's Ari Banerjee included) tend to side with Thiele, but Bias has some interesting points in favor of the commodity cloud...

(I was going to put up a short News Analysis comparing their statements... and I still might, but it's probably best to let the Microsoft/Skype fervor get out of the way first.)

paolo.franzoi 12/5/2012 | 5:05:39 PM
re: Clouds Need More Than SLAs


I think they are talking past each other, at least in the two videos.

The video presented here by Craig is talking about the way that essential IT functions are not provided by the Cloud.  The whole SLA thing as described here is important.  If you do not have the IT staff capable of dealing with it, then you will have to contract out that work.  If you need big scalable environments they can not be built without proper architecture and organization.  That is independent of cloud or not cloud.  If you are trying to do this without an IT staff then you are kidding yourself.

The link provided to Phil's video is more about how services are being built and what services are likely to be created effectively.  Which I agree with the idea that the lowest cost hardware is the way to go.  Dell putting Intel's design in a chassis is the same as Chang in Taiwan doing so.  There is no value (unless you want the service component) in a named vendor.  None.  The same holds true for Ethernet switches.  Basically all these things are reference designs dropped on PCBs.

These two views are not actually in conflict for me.  To think that somebody can take that commodity cloud and make an effective use of it without some rather skilled staff seems humorous to me.  Most small businesses can not do so unless they are doing some really simple things or buying off the shelf cloud services like Google Docs or Salesforce.com.

There was a point that I think was made but I will make it stronger than Craig's video.  Anytime you decide to put your application/service up in the cloud it better have the need to scale/tune based on load.  Otherwise, it is cheaper to run your own hardware in even a Tier 4 data center.  For my current application (email security) the cost to run this on AWS instances would be about 3x my current costs.

Finally, I completely disagree that the cheap and cheerful $50/month VM is not good for mission critical applications.  It is entirely excellent for those with a proper data architecture (which is where the persistence needs to be) and scaling/reliability architecture (which is how one protects the compute power).  The idea that some "special sauce" has to be in the cloud for this is silly.  It does link back to the entire point of Craig's guest about proper application design and architecture.



Sign In