◆ Home
◆ Recommendations
◆ PDF Resume
◆ Word DOC Resume
◆ LinkedIn
◆ STAR
|
| |
Wayne D. Roesner
15331 W 49th Ave ◆ Golden, CO 80403 ◆ (563) 505-3489 ◆ wayne@wayneroesner.com
|
|
| |
Initiative
Asserting one’s influence over events to achieve goals; self-starting rather than accepting passively; taking action to achieve goals beyond what is required; being proactive. “Tell me about a situation (personal, academic or professional) in which you were responsible for planning and organizing an event.” Hint: How did you get the assignment? How did you approach the task? How did you keep track of things? What tools did you use (to-do list, organizer, etc.) to help you? What was the first thing you did? What steps followed? How did you feel when the event took place?
|
|
Big Data Need at John Deere
|
|
Situation:
|
Big Data Need at John Deere
|
|
Task:
|
We have outgrown our databases. We need to research Big Data options.
|
|
Action:
|
I research Big Data options and landed on Hadoop. I then researched the area of Hadoop and its growing popularity. In my research I found there was an option for DIY (Apache Hadoop), MapR, Cloudera, Intel, and Hortonworks. I researched what Hadoop was and found it was really a bunch of products combined together on multiple computers. My first impression was to get the performance you need you had to purchased lots of hardware. I convinced a couple of friends to loan me some hardware and drives and I created a two node cluster in my office at home to understand it better. I then contacted the vendors to get pricing and what options were included in their solutions. At the time Cloudera was the market leader. I then worked with multiple departments at the IPN and HP (loaned us the equipment) to create a 10 node Hadoop cluster using Cloudera. We now had a playground for people to test out their theories.
|
|
Result:
|
Multiple departments and individuals throughout the organization tested Hadoop and its capabilities. After many successes it was decided to create a real Hadoop cluster on premise. This lead to a comparison of all Hadoop vendors and multiple hardware vendors. In the end it was decided to use Hortonworks and PSSC Labs Hardware (specific to Hadoop). There is now multiple 20+ node Hadoop clusters on premise. NOTE: I made the comment to purchase hardware has now changed. On a daily basis Deere is spinning up 300-500 nodes of Elastic MapReduce (EMR) at Amazon Web Services (AWS). Instead of spending $500,000 on a permanent 20 node on premise cluster, we now process data for $100’s of dollars per run. The on-premise cluster is still used today but they are now looking at using AWS. AWS is able to process in 1 day what would take months on premise at an extremely reduced rate.
|
|
|
|
|
| |
|
|
|