Can thin provisioning pay for itself?
This is my first post as a blogger at Ruptured Monkey so I hope I can live up the expectation of the quality that is apparent here.
I googled HDS USP V a few moments ago and found 32,800 related entries. What I was trying to find was more information on Thin Provisioning. From what I have found, I am still a bit unsure of what it means.
HDS' own brand of thin provisioning technology, Dynamic Provisioning software, lets customers allocate virtual disk storage based on their anticipated future needs without allotting physical disk storage up front. Should the need arise for more physical disks, customers can purchase more capacity later.
This practice saves businesses money because IT managers have traditionally over allocated storage capacity to avoid running out of storage and crashing applications. Dynamic Provisioning will also add more storage automatically, saving IT managers time because they don't have to spend time manually adding storage.
I would like to read this as making efficient use of my storage instead of wasting TB's of it on file systems that are only partially used.
Perhaps it could mean that I allocate 1 TB of storage to Server A, 5 TB of storage to Server B, 2 TB of storage to server C and finally 2 TB to Server D. That's 10 TB that I have actually allocated and spend good money on. Knowing the usage of most of our servers, I only need about 5 TB because the owners of Server B and C always ask for more than they need just to be on the safe side.
What I would like to see is HDS having 4 TB (a nice licensed size) sitting on the system in a Capacity on Demand type solution. Once Server B gets its act together, the storage would dynamically increase from 3 to 5 when it needs it but not go over that threshold. I would then put in an order to buy more storage which normally takes a week or two.
This type of provisioning becomes very efficient when you are dealing with 100 TB or more in a USP. If 40 % is wasted on file systems that are only partically full, that's 40 TB or lots of money. Just say, that equated to 1 million dollars, the new USP V will actually pay for itself.
The downside of this is ---- will it work? Can hosts actually increase their capacity dynamically? Where does the storage come from? How is contention for resources handled? How will already overworked array groups or ports cope with newly created LUN's?
I am really looking forward to an indepth explanation of what Thin Provisioning actually does with the USP V?
Oh, also where am I going to get 200 PB to see if the virtualisation works? I am still working on my first PB.