Check out Greg Shulz’s recommendations
on how to reduce your information footprint
whether it’s online,
offline or
nearline data.
Healthcare brings out a lot of passion in people these days. Many of the topics stridently debated are surprisingly relevant to enterprise storage:
Universal coverage
All business data isn’t created equal, but there is a threshold of storage, management, protection and security below which any business data should not fall. Sometimes people lose sight of where that line is, to their company’s and their career’s detriment.
Cost control
Data volume continues to grow faster than even my annual health benefits contribution. Businesses do the math – they just can’t increase their storage costs to match. Much like healthcare, one solution is to provide ‘good enough’ storage for low maintenance data and optional premium services for data that needs more performance and additional protection from risk. But how good is good enough?
Robin Harris has an opinion on this.
Who gets premium services?
Like healthcare, some of the most passionate debate is what data sets require that extra care. Storage technology is now starting to make these decisions automatically based on activity or other metrics. Still, mainstream IT struggles with knowing when to take data from the ’premium plan’ – mission-critical enterprise solutions – to slower, high-capacity options.
Who pays?
Healthcare costs are going up for everyone, and employees are carrying a heavier burden. Some say this incents people to make more efficient choices on medical costs. Data owners are seeing the IT version of a high deductible plan with more aggressive and granular chargeback policies.
Whatever the outcome of the Grand Healthcare Debate of 2009 and the less-momentus-but-still-important storage debate, we all benefit from the conversation.
What do you think?
“Server hugging” – the tendency of programmers to want their servers nearby like a warm security blanket – is not unique to Microsoft.
There’s a broader and more insidious ailment that affects IT organizations everywhere: can I sleep at night if my businesses’ data is out on the cloud? Let’s call it data hugging.
Data hugging is understandable. IT organizations are the stewards of a corporation’s digital crown jewels. It’s in their name, it’s why they exist.
With growth comes change, and IT needs to let go of their sentimental attachments to their data’s location. Cloud Computing is here to stay. It’s not for every byte and everyone, but it has a place in most organizations’ plans.
It’s time to set feelings aside and apply dispassionate business-based analysis to the in-house vs. Cloud analysis.
If you think you’ve got a lot of data, check out GigaOm’s look at some of the largest data giants ever. Facebook, for example, is expecting to store its 100 billionth photo this year.
How are they doing it? It’s about their software more than their hardware, because the software determines a lot about the physicial infrastructure required.
Facebook, Amazon and Google are all decoupling their systems from proprietary hardware. The extreme scale of their data storage makes it necessary to strip away anything that isn’t absolutely necessary for storing and recalling petabytes or exabytes of data safely and efficiently.
Does this trend foretell the future of enterprise storage in general? It’s not clear, but businesses have the same challenges on a smaller scale.
Is so, there is change and turmoil ahead for the established IT vendor world order. What else is new?
Energy efficiency is goodness, never more than this year. That’s reflected in the IT industry with a never-more-than-now focus on reducing power consumption. But some of the most valuable efficiency improvements are the ones that go beyond the obvious.
1. Do less. This is the obvious one. In storage, slower disk drives can be used that use less power. But much like scrapping your car for a bike, it works for only some situations and can reduce effectiveness productivity.
2. Do things more efficiently. For storage, this is the best bang for your buck. At the drive level it means more efficient mechanisms and power management techniques that reduce overall Watts/GB. For systems, it’s designing smart, turning of sectors and components when not in use, MAID, etc. These investments pay off every minute of every day, year after year.
3. Reuse power. Largely untapped in IT is the concept of reuse. IBM is innovating here by using supercomputer heat to heat the data center.
4. Reuse disk drives. Seagate is making data center disk drives re-usable with encrypted erase, which instantly wipes drives totally clean of their old data. Ten percent of enterprise disk drives reused would save close to a million drives from the scrap heap.
What’s number 5?
The EPA is working on an Energy Star specification for storage systems, much like they’ve developed for servers.
Unlike the server specification, the EPA is focusing on power consumption by storage systems as they read and write data, vs. while they are idle. Big mistake. Storage is mostly about…well, storing. The biggest impact comes from the steady-state power draw over the hours, days, months and years the storage system operates – not the relatively rare moments when data is coming in or going out.
That’s like measuring the energy efficiency of a refrigerator only when you open the door.
Storage read/write utilization time varies by application of course. But on average it is low. Maybe we need a Mega Data Center specification and a more mainstream specification.
What am I missing here? Why focus on active storage power consumption vs. idle?
Comments welcome.
The New York Times’ Tom Vanderbilt had an “inimical incuriosity” (now there’s some fancy New York vocabulary for you) about data centers before he wrote this thoughtful article. One can imagine his eyes bugging out as he is exposed to all the extreme technology coursing through the veins of these building/machines that power the data coursing through so many parts of our lives.
Highlights:
Thanks to Rich Miller at Data Center Knowledge for the heads-up on this.