It really amazes me that Cloud on z or the much more snappily titled zCloud isn't getting more press. Take for example:
US organisation Nationwide’s private cloud on IBM zEnterprise has replaced thousands of standalone servers,eliminating both capital and operational expenditure. The initial #consolidation exercise is estimated to have saved thecompany some $15 million over three years.
Check out the detail at http://ibm.co/13KWqeW
If you are doing Linux at scale (and by scale I mean more than 400 Virtual Machines) then the cost case for z is massively compelling. If you want to throw enterprise software into the mix from the likes of Oracle or even IBM's DB2 and WebSphere, then the license savings alone can pay for the underlying tin twofold within a year.
The IBM Enterprise Linux Server platform as an underlying platform for a private/public Linux cloud is a great message and doesn't come with the 'legacy' bagage oft associated (incorrectly IMHO but don't get me started...) of System z. For heavens sake it is a large virtualised server than runs Linux, how complex can that be.
As I have previously blogged, having a 'system of record' and then proliferating copies of this data to a myriad of distributed servers, just so you can do complex analytics against this data appears to me to be folly. When organisations are taking between 8 and 12 copies of their core data and squirting this via a multitude of various ETL methods out to other platforms and then having to manage these disparate systems, then the costs, not surprisingly spiral out of control. Not to mention the data management and consistency and operational decision making challenges this poses. Why oh why not just keep your data where it lives and breathes i.e. in your core system and analyse it from there? Surely this approach must be cheaper and not lead to a proliferation of data models...
Check out this Forester whitepaper for a banking discussion on this topic:
Link to the paper: http://ibm.co/Xm5P7y