Googlers Beta – 2009 Edition!
October 8th, 2009 | Published in Google Student Blog
We are relaunching our Googlers Beta Series and bringing you the unique stories and experiences of Google interns. Our first post for the season comes from Asit Mishra, a Google Summer Intern and PhD student from Penn State University, who will share with you his experience as a Software Engineer (SWE) Intern in our Seattle office.
Asit's Story:
Think about the 100+ tools that Google has launched, the billions who query our search engine everyday, and the 10,000+ crazy (in a good way) software engineers working at Google; there is a plethora of jobs that our server platform machines must serve each day. Our machines serve a diverse set of workloads, and managing and making sense of these workloads is a big challenge. But learning the behavior of these workloads can help us, among many other things, to make better scheduling decisions, to better project our machine growth, and to customize our cells to handle the jobs better. During my internship, I worked on a very interesting, and in my opinion, high-impact project where I modeled the workload of Google's machines.
I used a variety of tools, including time-series modeling, clustering and statistical methods to capture the workload signatures. After months of brainstorming with my mentor (Joseph Hellerstein) and the cluster management team, I was at a point where I could categorize the workloads into smaller groups. This gave us an opportunity to investigate the workload behavior at a macro-level and makes the analysis simpler.
Aside from the free food, massages and the rock-band set-up, I found the "Google-culture" to be really amazing. It is not very different from the culture one would see at a university or research lab. The openness and the helping attitude of Googlers, here, helped me a lot in my work. Between the numerous brainstorming sessions with my mentor and the cluster management team and the two trips down to Mountain View to dig-deep into our findings, these experiences have gone a long way in shaping the outcome of my project and enhancing my overall understanding of Computer Science. For me, it's been one of the most productive and enriching experiences within a 3-month span.
I had never seen terabytes worth of data before coming here! Once, my mentor told me that here at Google people don't worry about storing megabytes or gigabytes of data; if it's terabytes it a slight concern and if it's petabytes - that's where you actually start thinking.
I used a variety of tools, including time-series modeling, clustering and statistical methods to capture the workload signatures. After months of brainstorming with my mentor (Joseph Hellerstein) and the cluster management team, I was at a point where I could categorize the workloads into smaller groups. This gave us an opportunity to investigate the workload behavior at a macro-level and makes the analysis simpler.
Aside from the free food, massages and the rock-band set-up, I found the "Google-culture" to be really amazing. It is not very different from the culture one would see at a university or research lab. The openness and the helping attitude of Googlers, here, helped me a lot in my work. Between the numerous brainstorming sessions with my mentor and the cluster management team and the two trips down to Mountain View to dig-deep into our findings, these experiences have gone a long way in shaping the outcome of my project and enhancing my overall understanding of Computer Science. For me, it's been one of the most productive and enriching experiences within a 3-month span.
I had never seen terabytes worth of data before coming here! Once, my mentor told me that here at Google people don't worry about storing megabytes or gigabytes of data; if it's terabytes it a slight concern and if it's petabytes - that's where you actually start thinking.
A note from Joseph Hellerstein, Asit's host: We had great fun with Asit during the summer, and it was very productive. Asit's work is impacting how Google does cluster scheduling and capacity planning. Also, we submitted a paper to ACM SIGMETRICS Performance Evaluation Review. Oh, and we got some really cool jackets for our work with the Google Cluster Management Team.
Want more information on internships at Google? Visit http://www.google.com/jobs/intern for more information. You can also learn more about the exciting research going on at Google by visiting http://research.google.com/.