Some weeks ago I met Dave Merril, who’s involved in “storage economy” in HDS. Dave is a totally amazing guy who can open up your mind, just with a whiteboard and a little time. I warmly suggest everyone to follow his blog.

I already, many times indeed, wrote about the differencies between TCA and TCO, but I think it is still a prime time subject, and I’d like you to follow some reasonings about what I learned from Dave.

the price-tag

The price of storage hardware, I’m thinking of disks space and speed, cache size, connectivity, etc, is getting lower and lower. The average decrease rate of hard disks price is 15/20% per year! I’m pretty sure you can easily find a 1TB HD for less than 50 $ (a careful search could spot a cheaper one) 

The same reasoning as above can be done with enterprise storage. Prices decrease enough to allow you to buy much faster and capacious storage systems than just a couple of years ago, and I can see no reason why the trend of price versus performances will change for a foreseeable future. If you plot a graph of the storage price trend we’ll have a crystal clear depiction of what is happening. Any time we need some more storage we can find it at a lower price and we buy larger and larger quantities of it. 

Here is the matter in hand: pricetag and cost of the storage actually differ.

the cost of the storage

The price of the storage is just a minor component of the real cost of the storage! Operating costs trend is not a straight line, as often as not it is an upward slope, and these are the real costs that company has to incur.

When I refer to costs I mean everything (to cut a quite long list short): electrical power, datacenter floor, air-conditioning, support, human resources, backup and disaster recovery, helpdesk, planned (and not) downtimes, and so on…. Just think about what you need to do when the your new storage is finally delivered and you start planning a data migration: sometimes this can take months, months before you can dispose of your old storage, months when many operating costs are doubled!!!

It should be quite easy to understand that TCA (just the opposite of what usually happens) should be the least worry of an IT manager (or anyone in charge  of the purchase of the IT infrastructure).

The prices weigh a low burden when compared to the other costs related to IT infrastructure, especially when speaking about storage! The outcome of a simplified and conservative evaluation sets the ratio of all the costs over TCA to 4:1. I’m quite sure an assessment of actual cases leads to even higher ratios. 

bad in theory, likely worse in practice

Does storage costs remain the same? Surely not: if 5/6 years ago I bought a 10 TB storage, tomorrow I’ll likely buy a 30 TB one because of the badly increased capacity needs of my company. Bottom line: if the operating costs per TB was unchanged, now I’d have to cope with an expenditure roughly 3 times higher than before!!!. 

how to solve this problem

Storage needs will surely not calm down: data get more and more complex, data size is growing and people tend to keep a digital copy of every document without bothering of making physical copies out of every version. The only viable solution is to knock down operating costs during the whole life of the storage system. This can be accomplished in a couple of ways: either leveraging technology advancements or revising IT processes and methods.  Surely not every technology nor organizational model suites well and can fit in every company. A detailed analysis must be performed: how data are routed through the company, their birth, their archival, their lifespan and importance. 

Technology

New technologies developed by vendors usually come at a price (pun intended), an higher TCA, but the advantages given by the right new technologies are huge. 
We can see 4 main goals the technology innovation pursues:

  • easy of use
  • efficiency
  • automation
  • integration

Easy of use is indispensable to decrease the time needed to operate the storage and to employ less technically skilled personnel to perform everyday trivial tasks. Human resources always have an impact in IT budgets, and are usually limited. Having more effective tools is crucial to lower the hours per managed TB ratio.

Efficiency in a storage system is a very hot topic, often harshly debated by vendors, holding poor efficiency against each other. In short, efficiency can be measured as the capability to leverage at best the physical resources. I refer to “physical resources” because present-day storage are mostly built on a (or leverage an afterward added) virtualization layer enabling a whole new list of features. Most of these technologies/features are used to reduce data footprint (i.e. deduplication, compression), some improve the space usage (i.e. thin provisioning) some are related to performance optimization (i.e. wide striping). These technologies are not suitable for every environment and some storage systems pose constraints on which features combinations can be used.
By and large, the best option is the storage that can grant the highest operating freedom. The more efficient is the system, the cheaper will be its management. A very trivial example: if the space used by my data could be reduced by 50%, using a dedupe technology, I’d cut by half the disk space I need, and for sure power consumption and rack space usage!

Automation is just about the need to reduce human intervention while increasing the TB managed per system administrator. Among automation features “automated tiered storage” ranks first (I recently wrote about it), but many other innovative features are being offered to allow automating tasks that if otherwise performed by personnel would require time and higher costs. A couple more examples: automated NAS snapshot allowing users to recover files without needing the intervention of a sysadmin, automated unused space reclaim on hosts.

Integration with upper software layers (hypervisor, operating systems and applications) allows a quantum leap toward managing data instead of managing storage! If my application and the storage knows each other quite well, you can leverage the storage when in need, say, to clone a big DB in a bunch of clicks and minutes, or to copy a whole ERP production environment to let the developers do their job! This kind of integrations have the advantage of speeding up IT department responsiveness thus simplifying and relieving other departments jobs. Bottom line: overall increase in efficiency and productivity of the whole company.

methods, processes and reorganization

The technological part is easy to comprehend (and put in practise), I cannot say the same about cost reduction measures that would come from reorganizing and honing processes and methods. 

It is really difficult to have a hand in changing the way data originate, spread and are misused by final users! Just think about a corporate memo sent to every employee via email, along with its huge attached document. Most of the personnel will surely copy the attachment on his/her own private area carved in the company-shared storage, without even reading it. It is just as simple as frequently witnessed. To prevent the misuse of IT resources in your company (not just thinking about storage), a firm educational activity must be enforced, as well as establishing a simple, shared set of rules that any user should obey.
I can almost hear the the burst of laughter (often turning bitter) most IT manager have thinking about suggesting the company owners such radical ideas. 

We know that the ‘better safe than sorry’ approach eventually pays off, for sure, but it is seldom carried out. Anyway one approach does not rule out the other, the sum of technology and right policies would noticeably knock down storage overall cost.