It was in Las Vegas. It coincided with the Las Vegas Rock ‘n’ Roll marathon on the strip and the professional rodeo associationNational Finals Rodeo starts as Gartner ends. Gartner is a bit of a marathon for three and one-half days and there is some bull to watch out for, but overall, it's a worthwhile event.
In no particular order and in Tweet style, here are tidbits that struck me from this year’s conference. [I’ve added some of my thoughts in brackets.]
Overheard from a security guard when I said I was looking for someone at lunch: “That should be easy if they are white, over 40 and have greying hair.” Ta Dum
Conversation with a customer: “Of course we only use Brocade for our SAN. And until this new project, we only used Cisco for the network. But we are going to look at other network vendors. It’s time.”
To the Business, it’s not about how big the Data is, it’s about gaining a Bigger Context and thereby increasing awareness about customers and markets at a larger scale.
Glasses Law: For every 25% increase in feature/function, there is about 100% increase in complexity. [Simplicity is increasingly critical to datacenter architecture.]
To paraphrase Capitan Sully Sullenberger: It was 200 seconds that would define my and every passenger’s life. I’m asked what did I do when I realized we had lost both engines? Well, three things: I stopped being afraid as quickly as I could; I focused on the most important things to do next and ignored the rest; I kept my co-pilot in the loop and we verified that we had done everything we could think of before we touched down on the water. [Sound advice when the next crisis comes your way.]
Current market analysis suggests 90% of enterprises will bypass Windows 8. This is supported by only 21% completing Window 7 upgrade and 34% who haven’t even started yet [Left unsaid, is what does Microsoft do about that?]
Big Data projects are not stalled due to technology; they are stalled due to lack of brain tissue. Where are the data scientists? Estimate that four million unfilled openings for them will exist by 2015.
Infrastructure & Operations (I&O) are spending more than 40% of the time keeping the lights on, i.e., just running what they already have. [Complexity prevents them from having time to innovate and anticipate, reinforcing the notion they aren’t adding value to their business.]
Data center projects (new and expanding existing facilities) are planned by 50% of customers for next 2 years. Why? Lack of power, lack of space.
Scale-out is the architecture for server, storage and networks to achieve lower cost at larger scale. This requires 10 GE networking to get started.
Solid State Disk is not rotating rust. But, most SSD deployments assume this is the case. Next generation will fully exploit the difference between SSD and HDD behavior.
Dark Data is like Dark Matter. It’s about 80% of what you are storing, but no one can see it. Many Big Data projects are targeting existing data to find new value.
Data center voltage to the rack is moving to 440v to keep up with the power demand of dense compute and storage clusters.
The new model for Big Data analytics is to move executable modules to the data. [This inverts the client/server model for the IO subsystem and affects the application architecture.]
Texas Advanced Computing Center Stampede has 500,000 cores, uses 40 KW/rack, has 200 racks, stores 10 PB of data and can support 10 PFlop of floating point computation. Some research problems consume all this resource when they run!!!
SSD from the consumer market is going into public cloud data centers since it’s one-third the price of enterprise grade.
Venture capital invested $700 million so far in about 80 SSD startups. By 2015, 80% of these will disappear (died or bought-out).
Award for highest infomercial content to time spent listening. Cisco. [Many tweets of dissatisfaction were flying as chest thumping became embarrasing.]
Interesting vendors I chatted with: Virsto Software and IXIA. Informed people with interesting solutions.
The infrastructure is an application. [I think this is the principle underlying software defined networking and other trends that simplify datacenter operations.]
Value = Why / How. First, focus on Why to get the most value from your time and energy. It will force you to find (and kill) orthodoxy and dogma which are the cause of high cost and eventual obsolescence.
Systems thinkers are critical to any team. You can’t succeed with just technology experts. Systems thinkers envision the new and can visuallize the cascading impacts of point changes in your data center. [Get some.]
The internet of things talking to each other will dwarf the internet of people jabbering on Twitter.
How to be happy in your work: Autonomy, Mastery, and Purpose. [“Work Hell” is the opposite. Why not build a “heavenly” team?]
People who attend Gartner DC Conference like to talk to each other. People always introduce themselves when sitting down at breakfast and lunch and welcome you to join in their conversation. [IMHO, this is not the culture at larger shows like VM World, EMC World, etc.]