Cheaper To Buy vs Cheaper To Own: Hypervisor Economics Pt.1
The data explosion, of late, has been phenomenal. We haven’t even begun to harness and optimise all the data generated by traditional systems, and now we have to contend with high-velocity and high-volumed Big Data.
The good news is that the price of storage has come down, thanks to Moore’s Law.
But that’s the price of acquiring storage. Do we simply keep acquiring more and more storage as data grows by leaps and bounds? Can we ever hope to keep up?
There have been technologies touted; the foremost of which is hypervisors; that proposes we use the hardware we already have in our data centres and reduce if not cut out the cost of acquiring more hardware. That is a very good idea, given that these lean economic times have seen many IT departments scrambling to do more with less.
But how are they doing more with less? Is the ‘more’ enough? Is the ‘less’ enough?
It stands to reason that every initiative to reduce something should first start with ensuring that, that ‘something’ can be quantified. For Hitachi Data Systems (HDS) Malaysia managing director, Johnson Khoo, customers are going to, If not already, face the very important question of how to manage the costs of non-physical and virtualised infrastructure. “Managing the physical hardware is already tough enough, what now with non-physical infrastructures like virtual machines (VMs)?” he pointed out.
Hypervisors is an interesting technology as it proposes to save cost. Found in test and development environments just some years before, Merrill observed that now more and more hypervisors are in production workload mode, there are problems with getting them to scale. Jaw-dropping price methodologies by the incumbent from among just a handful of viable virtualisation vendors, also draws attention to the fact that an organisation’s IT could face a vendor lock-in. There weren’t viable alternatives to VMware before, and even though observers view that there is pressure to use multiple hypervisors and 2012 is the year of Hypervisor Heterogeneity, a good number of data centres are already taken hostage at the systems management stack level. The need to save becomes more urgent.
Merrill: Would you pay more if it could save you more?
The Hitachi Data Systems (HDS) view is that the price of a piece of technology isn’t necessarily what it’s going to cost you, or the total cost of ownership (TCO). Chief economist David Merrill stated, “Price does not equal cost.” Considering how the price of storage of has been dropping thanks to Moore’s Law, the cost of acquiring storage has gone down (and is still going down) from 60-percent of TCO to about 15-percent of TCO today. What does the remaining 85-percent comprise of?
The storage vendor recently hosted a roundtable together with Enterprise IT News entitled, “The Dollars and Sense of Hypervisor Economics” and what transpired was an interactive discussion about all the unexpected elements that can contribute to the cost (reduction) of storage.
Raise your hand if you want to cut costs!
That is a statement that will have immediate hand-raising to, followed by a just-as-swift “How??” Since 2002 when HDS’s economics consultants officially came up with methodologies, they have documented and characterised some 34 different types of costs that make up the total cost of owning storage.
Labour, maintenance, power and cooling currently have been found to cost some three to four times more than cost of acquiring storage.
The good news is that there are HDS-proven strategic and tactical investments for storage infrastructure cost reduction, and they can range from technology components to organisation and staffing changes, and even storage provisioning methods to the end user.
“When we start to understand which types of costs are important (that you want to reduce), then you can start to correlate exact technologies and options that are available to reduce costs,” said Merrill.
In fact according to him, the three primary techniques of reducing costs are virtualisation, dynamic tiering and over-provisioning/thin provisioning.
When it comes to the cost of hypervisors or virtual machines (VM), about 24 identified costs for storage apply for VMs as well. Merrill said, “Hypervisors have the same kind of framework as storage in terms of that the cost of it is more than just the price of the VM license.”
Once we understand the total cost of things, and all these cost elements, then we can start to make ideas to go reduce those costs. “
A well-worn OXYMORON: Pay more to save more
Khoo observed it was interesting how storage has gained importance over the past two years. “Organisations are starting to see it as a key cost element especially with the increasing popularity of social networking sites. We have not talked about storage so often and so regularly as we have the past two years. If we are not careful (with how we handle and provision storage), it is going to choke up an organisation.” All that new data we are generating has to go somewhere to be stored, after all.
That well-researched figure that IT solutions vendors used to toss around when peddling their wares; 30- to 40-percent of IT budgets going into maintenance and just keeping the lights on; begins to creep to mind. Merrill opined, “It wouldn’t be uncommon for 40-percent of your IT budget to go towards storage-related costs.”
So, it also starts to become clear that asking for price discounts on hardware, doesn’t really mean much, if in the longer run, you start to incur total costs that exceed the price of acquisition, or the TCO of another hardware solution. Merrill also gave cases of how the cost of electricity to power storage arrays and cooling started to cost more than the storage array itself after four years.
“Would you pay more if it would save you costs in the long run?” That was one of the more important questions the chief economist raised during the roundtable.
If organisations really want the cost of storage or TCO to go down, they have to look at much more than just the price. Technology options achieve different things. There are some storage architectures that suit high-growth environment, others reduce compliance or legal costs and yet others control power and space costs.
(Coming in Part 2: The structured design and cost planning approach)