Vinayak Hegde

Recently in the last month, I traveled to Bhutan to attend the SANOG Conference. Bhutan is a small country nestled in the Himalayan mountains surrounded by Indian to the East, West and South and China to the North. It was a good opportunity to meet some like-minded network geeks and also visit an exotic country.

SANOG (South Asian Network Operators Group) is a conference where various stakeholders from the Internet infrastructure ecosystem can come together, share operational experiences and learn from each other. SANOG is targeted towards the SAARC Countries (India, Pakistan, Nepal, Sri Lanka, Bangladesh, Bhutan and Maldives). SANOG is loosely modeled on APRICOT conference with 5 days of Workshops, 2 days of tutorials and 2 days of conference.

The 16th edition of SANOG was held in Paro, Bhutan. This was the second time that Bhutan was hosting SANOG. SANOG was held in the the Paro Engineering College beside the Paro river in very picturesque settings. The earlier edition of SANOG in Bhutan was held in the capital, Thimphu. This was the 2nd time I was attending SANOG, having attended an earlier edition of SANOG in Mumbai in 2006.

I attended the workshop on Network Security by Gaurab Raj Upadhyay and Johhny Martin from PCH. It covered the basics of security specifically for ISPs and large network providers. There were some good discussions on how to manage the different security audit process as well as an incident management program in case of network security breaches. The hands-on part of the workshop concentrated heavily on securing backbone routers and exchanging routes information securely. Some aspects of filtering and verifying network traffic were also covered. The last day had demos of several tools such as nessus and nmap. The slides can be downloaded from the SANOG Program page. Also in between the workshop breaks and during one of the days of the workshop, Devdas and I wrote a improved whois server that is hopefully in production now at the NIC website.

In the tutorials part of SANOG, I was giving a half-day tutorial on application-level performance measurement [Slides,PDF]. There was an small but interested crowd in the tutorials. I ended up covering a lot more of the web-facing and measurement tools as many of the participants were application developers who had written quite a bit of PHP code. It was the first time I was giving a tutorial on this topic and it helped that it was interactive. In addition to the material on the slides, I talked a bit about front-end performance and tools such as Yslow (Yahoo), Pagespeed (Google) and Webpage Test (AOL). There was a lot of whiteboarding and veered a little away from the slides. I also spoke about the network measurement work being done in the IPPM, BMWG and PMOL working groups in the IETF. The feedback was pretty good and I plan to give a longer version tutorial at later versions of SANOG/APRICOT. I skipped the second day of the tutorials and went to Chele La pass.

The conference had several talks that I was looking forward to and I was not disappointed. The standout talk were on long distance wireless network deployment by Matt Peterson and F-root update by Pete Losher. Both has interesting networking insights and interesting traffic data. I also gave a talk on my IETF fellowship experience [Slides,PDF]. Some of the slides were liberally lifted from “The Tao of the IETF” written by Paul Hoffman. As (good) luck might have have it, I ran into Paul Hoffman at the IETF 78 and told him about it . There were a few questions about the fellowship after the conference so I hope it would inspire more people to apply to the IETF fellowship.

The 2nd CloudCamp Bangalore was held at Dayanand sagar College of Engineering. It was co-located with the First Hadoop summit in India. The Hadoop summit was interesting and more relevant to me as I am using a Hadoop cluster for Analytics at Inmobi. Dave kicked off Cloudcamp with signature “unPanel”. I was on the Unpanel this time and answered some questions on mobiles, netbooks and smartphones as access devices for the cloud and the on impact of Google patent on MapReduce.

The corridor discussions with a bunch of Hadoop committers were insightful. I also found out more about Mahout. Mahout is a Apache project to build scalable machine learning libraries. It is not restricted to Hadoop implementations, but much of the current activity seems to be around Hadoop.

Notes and embedded slides from the sessions I attended follow:

Hadoop summit Keynote

Data Management on Grid

Notes:

• Y! uses a HDFS replication factor of 3 (the hadoop default) in most cases. Exceptions are big clusters with large number of applications running simultaneously.
• Y! does not use Avro yet due to large amount of legacy data. Twitter uses Avro.
• Data ingestion layer uses MapReduce for heavy lifting and format conversion for storage.
• LZO is used for compression. gzip (not ideal due to non-block-level indexing) and bzip2 is also used. There are problems with slowness of bzip2 decompression but bzip2 delivers better compression ratios.
• Data ingestion layer also oversees policy for data retention and purging.
• Underlying filesystems is rarely a bottleneck for Hadoop. Mostly the synchronization semantics of HDFS is a bottleneck. A file operation is not successful until all the replicas are in sync.

Machine Learning using Hadoop

Notes:

• There are clear differences between data mining and machine learning.
• ML is harder to implement efficiently on Hadoop. Improving efficiency is still a research problem.
• Hadoop creates one map job / block creating too many empty files and also many reducers.

Optimizing and Benchmarking Hadoop

Notes:

• As a thumb rule, adding as much memory as money can buy is a a good idea for Hadoop
• Consider Network connections as shuffle stage does heavy network I/O
• Solid state disks might make sense at certain price/performance ratios. They are also more power efficient.

Tuning Hadoop To Deliver Performance To Your Application

Notes:

• Several parameters to tune Hadoop but must be used in conjunction with each other.
• Set number of map jobs slightly more than number of cores to ensure better utilization. Makes sure that data is processed in waves. Also better network utilization (as shuffle phase happens parallely with Map phase) along with CPU scheduling
• Choosing a good HDFS block size is important. Number of HDFS blocks is directly proportional to number of Map tasks generated

Links to all presentations

• Hadoop Summit Keynote – USage inside Yahoo and Statistics
• Marketing presentation – Yahoo’s committment to Hadoop and Opensource
• Data Management on Grid
• Machine Learning using Hadoop
• Multiple Sequence Alignment Using Hadoop
• Optimizing and Benchmarking Hadoop
• Challenges And Uniqueness Of Quality Engg And Release Engg Processes In Hadoop
• Tuning Hadoop To Deliver Performance To Your Application

We recently had our flagship event Headstart 2010, Hyderabad . This was the fourth edition of Headstart and the first one in Hyderabad. The previous editions were held in Bangalore and Mumbai. The event was co-located with ITsAP (HYSEA) at the HICC Hotel in Hi-Tech City. We had a good response from the startups.

This time we had a similar format as the previous Headstart. The on-stage demo and demo-pit were provided free of cost to the startups (We only charged for registration of the startups). The startups made good use of the demo-pit space. A few startups generated quite a bit of buzz including Notion-ink (makers of the Adam Tablet PC), Robosoft Systems (Makers of duct cleaning robot), Sysblitz (on-demand Business applications), Youpid (Tapping into the marriage market using social networks), Go-live gaming (Educational products using gaming) and Innoz (A student startup from Kerala that makes mobile applications for filesharing).

The complete list of selected startups

Two of the panels “Unconventional Sources of funding” and “Business Plan mistakes” were well received by packed halls. The questions posed to the panelists and the conversations with the attendees during the conference pointed to the fact that the startup scene in Hyderabad is still nascent and needs to mature as compared to Mumbai and Bangalore. We were happy to help connect many people in Hyderabad through the Headstart event and our regular Startup Saturdays there.

More coverage

• On Twitter
• On Swaroop’s Blog – one of the panelists on “Unconventional sources of funding”
• On the Youpid Blog – one of the startups in the demo-pit

ACM Compute 2010 concluded yesterday. It is the flagship conference of the ACM Bangalore chapter. This year was the 3rd edition of the conference and more than 500 people attended the conference. The highlight of this year’s conference was the launch of ACM India. ACM wants to increase it reach in India and ACM India Council consisting of 18 leading computer scientists from academia and industry are heading this initiative.

The ACM India launch was addressed by 3 Turing Award Winners – Barbara Liskov, C.A.R Hoare (Tony Hoare) and Raj Reddy. The ACM Turing award is “The Nobel Prize for Computing” and it is rare to see three Turing Award winners address the audience at any event. Barbara Liskov is the most recent awardee of the Turing award (the 2nd woman to win it) and she spoke on the power of abstraction. She spoke about the problems early programmers faced when writing large and complex programs. She explained how she tried to solve it using abstractions similar to (what is now called) Object-oriented programming. She talked at length on how her insights and experiences with these programming problem led to design of the CLU language. CLU was the first language to implement iterators and generators (as well as exception handling). It was a good lesson in computer history listening to her. I learned later that she was the first woman to get her PhD from a Computer Science Department. (Her doctoral advisor was the legendary John McCarthy). Her presentation and the mentioned references in it make for good reading.

Dr Raj Reddy is the only Indian who has won the Turing award for his contributions to field of Artificial Intelligence. Incidentally, his PhD advisor was also John McCarthy – AI Pioneer and Turing Award winner. Dr Raj Reddy spoke about the growth of computing over the years and the challenges of reaching the “bottom of the Pyramid”. He explained why there was need to move from the WIMP-paradigm in user interfaces to the SILK (Speech, Image, Language and Knowledge) to increase the reach of computing. His Turing award lecture (“To dream the possible dream”) makes for interesting read as well.

C.A.R Hoare (Tony Hoare) was the next speaker. He is a living legend in computer science. I was looking forward to hearing him speak as I had studied the Quicksort algorithm (which he invented) and Communicating Sequential Processes paper in college. He was remarkably witty and his enthusiasm for computer science shone through in his talk. In particular he spoke about the Verified Software initiative which he contended was similar in scope and impact (for Computer Science) to the Hubble Telescope and the Human genome project.

The following 2 days, we had the ACM Compute 2010 conference and there were several hands-on Tutorials on Cloud Computing, Rich Internet Applications and Web 2.0 apps, Widgets and Mobile Applications. The RIA tutorial was conducted by Mrinal Wadhwa (slides embedded below) and the Facebook connect tutorial by Prateek Dayal (of Muziboo).

(Disclosure:I am the secretary of the Bangalore Chapter and am on the program committee for ACM Compute 2010.)

With winter approaching fast we are also into conference season.

ACM Compute 2010 – 22nd & 23rd Jan, 2010.

ACM Bangalore chapter is organising ACM Compute 2010 which is into it’s third year now. This year the broad theme is Cloud computing and Information retrieval, management and analytics. The aim of this conference is to bring together researchers, practitioners, technology market movers, and thought leaders, with a view to advance the state of the art, and the state of the practice in applied research. This year we are planning to do something special – details soon The Call of papers (CFP) is out for sometime now and the last date for submissions is Oct 1 2009. You can also submit a proposal for a half-day or day long tutorial. Last year we had a bunch of good tutorials and also the symposium on Cloud Computing co-located with ACM Compute 2009 which was a great success.

Disclaimer: I am on the program committee of ACM Bangalore and am the secretary of the ACM Bangalore chapter.

Pycon India 2009 – 26th & 27th Sep 2009

Also this weekend (26th and 27th September 2009) India’s first Pycon India 2009 is being held in Bangalore. There is an interesting list of talks lined up. So do register if you are interested in attending.

Foss.in – December 1-5, 2009.

Foss.in shifts to a new venue this year NIMHANS convention centre. This year promises to be interesting as the venue is available for longer durations. Also there are going to be hacker evening/nights where tinkerers can meetup and talk about a whole range of stuff not restricted to just FOSS. My educated guess is something on the lines of CCC in Germany. Definitely something to look forward to. Plus I think there will be atleast one evening where we will have music . So join the mailing list if you are interested in presenting/attending as more details should emerge soon.