Dharmesh Kakadia

News : My book Apache Mesos Essentials is now available !

Welcome to my page! You can find me as @dharmeshkakadia on most other places. My email is dharmesh.kakadia@research.iiit.ac.in. My resume is here.


2013 - now

Research Fellow @ Microsoft Research

Nov, 2013 to now

I am working with Dr. Kaushik Rajan on PerfOrator project. I am building ways of constructing critical path models for Yarn, Hive and Hadoop jobs from fine grained statistics. The goal is to predict execution time and resource requirements of jobs. We plan to integrate it with reservation based scheduler. I played various roles including intern and consultant within the team.

2011 - 2013

Research Assistant @ SIEL

from December, 2011 to November, 2013

Cloud Computing Group

I am working with Prof. Vasudeva Varma as part of cloud computing group at SIEL. I mainly focus on scheduling related problems in large scale distributed deployments.

Summer, 2013

Google Summer of Code with Apache CloudStack

from May to Sep, 2013

Integrating Apache Mesos with CloudStack through cloudformation service. You can check the documentation

Summer, 2012

Intern @ Unisys corp.

from May to Aug 2012

I worked with Mr. Nandish Kopri on Unisys clearpath system. My focus was on bringing network awareness to the hardware partition scheduler. Developed a method for identifying reconfiguration of partitions to improving network scalability and application performance.

2011 - 2010

Intern @ Tata Research Development and Design Centre

from December, 2010 to April, 2011

Embedded Systems Software Research Group

I worked under mentorship of R Venkatesh (aka venky) and Ulka Shrotri to solve the problems in formal analysis of embedded softwares. Addressed challenges in portability and scalability of the in house automated test data generation tool.

Early, 2013

Intern @ Doctrz

from January to April, 2013

I was responsible for design and implementation of the back-end platform for the healthcare startup. Challenges involved are data encryption, combination of relational and NoSql database and offline access to the platform from multiple devices.


2011 - 2014

International Institute of Information Technology, Hyderabad

from 2011 to 2014

MS by Research (Cloud Computing) (CGPA : 9.5/10)

2007 - 2011

Dharmsinh Desai University, Nadiad

from 2007 to 2011

Bachelors of Engineering (Computer) (Percentage : 74.5/100)


Performance models for SLA Aware cluster management

Dharmesh Kakadia, Kaushik Rajan, Carlo Curino and Subramaniam Krishnan

MSR-Tech Report.

Apache Mesos Essentials

Wrote a book with Packt Publishing about Apache Mesos focusing on practical aspects. It covers running various frameworks (Hadoop, Storm, Chronos, Marathon), Mesos internals, writing Mesos applications and administering Mesos clusters. Its available on O'Reilly and Amazon as well.

Network Virtualization and Cloud Computing

Dharmesh Kakadia

Winner of Cloud 20/20 v3 student technical paper contest, organized by Unisys Corporation.

Energy Efficient Data Center Networks - A SDN based approach

Dharmesh Kakadia and Vasudeva Varma

IBM Collaborative Academia Research Exchange (I-CARE) 2012.

Optimizing Partition Placement in ClearPath Forward!

Dharmesh Kakadia and Nandish Kopri

White-paper and Patent US20140136688

MECCA: Mobile, Efficient Cloud Computing Workload Adoption Framework using Scheduler Customization and Workload Migration Decisions

Dharmesh Kakadia, Prasad Saripalli and Vasudeva Varma

ACM Mobile Cloud'13

Network Virtualization Platform for Hybrid Cloud

Dharmesh Kakadia and Vasudeva Varma

Network Infrastructure Services as part of Cloud Computing Workshop at IEEE CloudCom’13

Network-aware Virtual Machine Consolidation for Large Data Centers

Dharmesh Kakadia, Nandish Kopri and Vasudeva Varma

Network-aware Data Management Workshop at SuperComputing'13

Seed Selection for Domain-Specifc Search

Ajay Dubey, Nikhil Priyatam Pattisapu, Krish Perumal, Sai Praneeeth, Dharmesh Kakadia and Vasudeva Varma

Vertical Search Relevance Workshop at WWW‘14

Measuring Diversity of a Domain-Specific Crawl

Nikhil Priyatam, Ajay Dubey, Krish Perumal, Dharmesh Kakadia and Vasudeva Varma.


Virtualization vs Containerization to support PaaS

Rajdeep Dua, ReddyRaja Annareddy and Dharmesh Kakadia

IEEE International Workshop on the Future of PaaS at International Conference on Cloud Engineering (IC2E) 2014

Energy and SLA aware dynamic scheduling and provisioning algorithms for Cloud Data Centers

Implemented virtual machine provisioning and scheduling algorithms for virtualized‎ data center environment. The algorithms minimize the energy consumption and SLA violation using past utilization history.




Run big data platforms over multiple cloud. The system aims auto-scaling capabilities and intelligent scheduler to optimize job completion time and price, exploiting features like spot instances and deadline prediction. Currently supports Hadoop and Spark over AWS and OpenStack.

Enhancements to Automated Test‐data Generation tool

Proposed and implemented the solutions for the portability, scalability and efficiency issues of AutoGen, an Automated MC/DC Test‐data Generation tool for C programs, at TRDDC. The tool is based on the Formal Analysis of code. The implementation included building Common Functions of Standard C Library and POSIX Standard for model‐checking, developing an unparser to convert Intermediate Representation (IR) to C code and the development of Program Slicer.

Implementation of Compiler for subset of C

Built a compiler and runtime for a small language. The language has all standard features including variables, constants, arithmetic, logical and relational operators, looping, program control, functions and terminal I/O functionalities. The runtime was stack machine based interpretation of the generated Intermediate code.

Review Summarization System

Summarizing the user reviews for various products from various sources, so as to provide accurate and collective feedback about products to the manufacture and other potential users. The project included preprocessing reviews, identifying products and features, quantifying the satisfaction/dissatisfaction of the user from review and presenting the most relevant reviews on the user query.

Mini Search Engine for Wikipedia

Implemented a mini search engine for Wikipedia data. This project included building index subsystem and search functionalities. All the major phases of Information Retrieval and Extraction, from preprocessing, indexing, ranking and answering user queries were built from scratch for to provide highly efficient indexing mechanism and sub second response time.

Set Expansion using Contextual Information

Developed a generic set expansion system. Given a seed list of items from the same concept set, the idea is to rely on semi structured web pages and generate the complete set using some of the contextual information present on the web. The expanded set can be used in a variety of ways from giving better search results to content recommendation.

Privacy Preserving Sharing on Cloud

Worked on combining techniques for data privacy and privacy preserving computation to give the most query results in lesser complexity while revealing minimum information.