Anshul Gandhi

  


      
Anshul Gandhi

Assistant Professor
Department of Computer Science
Stony Brook University

Email: anshul (at) cs (dot) stonybrook (dot) edu

CV: [pdf] (updated: May 2014)

[News]   [Background]   [Research Interests]   [Publications]   [Patents]   [Teaching]   [Service]

News:

[Top]

I will be starting at CS@SBU this Fall (2014), and plan to recruit a couple of Ph.D. students.

I am broadly interested in systems and performance modeling. My research aims to improve system performance via analytical tools.
My recent projects include cost and performance optimization for cloud applications (ongoing, with IBM) and power management in data centers (PhD thesis).
Please see my research interests and publications below for more information.

Feel free to email me if you want to learn more about my research and are interested in joining my group.

Background:

[Top]

I am currently a Post Doctoral Researcher in the Cloud Optimization and Analytics group at the IBM T.J. Watson Research Center.
Starting August 2014, I will be an Assistant Professor in the Department of Computer Science at Stony Brook University.

I completed my Ph.D. in 2013 from the Computer Science Department at Carnegie Mellon University where I was advised by Prof. Mor Harchol-Balter.
You can find my thesis here - Dynamic Server Provisioning for Data Center Power Management.
I was awarded the 2013 SPEC Distinguished Dissertation Award for my thesis.

I completed my undergraduate studies in 2007 from the Department of Computer Science and Engineering at Indian Institute of Technology, Kanpur.

Research Interests:

[Top]

I am interested in applying theory to systems. In particular, my research aims to leverage mathematical tools such as {Performance Modeling,Queueing Theory,Control Theory} to analyze the behavior of systems such as {Distributed Systems,Cloud,Data Centers} and to optimize metrics of interest such as {Performance,Power,Energy}.

I am currently working with IBM Research on applying queueing theory and control theory to provide performance guarantees for cloud-deployed applications (via scale-out and scale-up in OpenStack).

Some of my recent projects include:
 Providing performance guarantees for cloud applications (ICAC '14)
 Scaling multi-tier applications in response to unpredictable workload demand (TOCS '12, Performance '10 (a), OpenCirrus Summit '11)
 Dynamically provisioning the caching tier (ICPE '14, Middleware '12, HotCloud '12)
 A hybrid (predictive-reactive) resource provisioning approach for data centers with partial workload forecast (Sustainable Computing '12, IGCC '11)
 Analysis of multi-server systems with setup costs (provisioning/boot times) (QUESTA '14, Sigmetrics '13, OR Letters '13, Performance '10 (b), MCQT '10, MAMA '10)
 Sensitivity analysis of sleep states in servers (IGCC '12, HotPower '11)
 Impact of scaling (size of the data center) on performance and power consumption (Allerton '11)
 Optimal allocation of data center power budget among servers (Sigmetrics '09)

Publications:      [Google Scholar]  [DBLP]  [ResearchGate]

[Top]

  Adaptive, Model-driven Autoscaling for Cloud Applications
Anshul Gandhi, Parijat Dube, Alexei Karve, Andrzej Kochut, Li Zhang
ICAC 2014 [pdf]

  Exact analysis of the M/M/k/setup class of Markov chains via recursive renewal reward
Anshul Gandhi, Sherwin Doroudi, Mor Harchol-Balter and Alan Scheller-Wolf
Queueing Systems, Volume 77, Issue 2, pp. 177-209 [pdf]

  Dynamic Management of Caching Tiers
Anshul Gandhi
ICPE 2014 [pdf]    (invited abstract)

  Exact Analysis of the M/M/k/setup Class of Markov Chains via Recursive Renewal Reward
Anshul Gandhi, Sherwin Doroudi, Mor Harchol-Balter and Alan Scheller-Wolf
SIGMETRICS 2013 [pdf]    (also, Technical Report CMU-CS-13-105 [pdf])

  M/G/k with Staggered Setup
Anshul Gandhi and Mor Harchol-Balter
OR Letters, Volume 41, Issue 4, pp. 317-320 [pdf]    (extended version, Technical Report CMU-CS-09-166 [pdf])

  AutoScale: Dynamic, Robust Capacity Management for Multi-Tier Data Centers
Anshul Gandhi, Mor Harchol-Balter, Ram Raghunathan and Michael Kozuch
Transactions on Computer Systems, Volume 30, Issue 4, Article 14 [pdf]    (also, Technical Report CMU-CS-12-109 [pdf])

  SOFTScale: Stealing Opportunistically For Transient Scaling
Anshul Gandhi, Timothy Zhu, Mor Harchol-Balter and Michael Kozuch
Middleware 2012 [pdf]    (extended version, Technical Report CMU-CS-12-111 [pdf])

  Saving Cash by Using Less Cache
Timothy Zhu, Anshul Gandhi, Mor Harchol-Balter and Michael Kozuch
HotCloud 2012 [pdf]

  Are sleep states effective in data centers?
Anshul Gandhi, Mor Harchol-Balter and Michael Kozuch
IGCC 2012 [pdf]

  Hybrid Resource Provisioning for Minimizing Data Center SLA Violations and Power Consumption
Anshul Gandhi, Yuan Chen, Daniel Gmach, Martin Arlitt and Manish Marwah
Sustainable Computing, Informatics and Systems, Volume 2, Issue 2, pp. 91-104 [pdf]

  The case for sleep states in servers
Anshul Gandhi, Mor Harchol-Balter and Michael Kozuch
HotPower 2011 [pdf]

  Minimizing Data Center SLA Violations and Power Consumption via Hybrid Resource Provisioning
Anshul Gandhi, Yuan Chen, Daniel Gmach, Martin Arlitt and Manish Marwah
IGCC 2011 [pdf]    (Best Paper Award)
(Pick of the Month for March 2012 in the IEEE STC on Sustainable Computing)

  Distributed, Robust Auto-Scaling Policies for Power Management in Compute Intensive Server Farms
Anshul Gandhi, Mor Harchol-Balter, Ram Raghunathan and Michael Kozuch
Open Cirrus Summit 2011 [pdf]

  How Data Center Size Impacts the Effectiveness of Dynamic Power Management
Anshul Gandhi and Mor Harchol-Balter
49th Annual Allerton Conference on Communication, Control, and Computing [pdf]

  Server farms with setup costs
Anshul Gandhi, Mor Harchol-Balter and Ivo Adan
PERFORMANCE 2010 [pdf]
Performance Evaluation, Volume 67, Issue 11, pp. 1123-1138

  Optimality Analysis of Energy-Performance Trade-off for Server Farm Management
Anshul Gandhi, Varun Gupta, Mor Harchol-Balter and Michael Kozuch
PERFORMANCE 2010 [pdf]    (extended version, Technical Report CMU-CS-10-108 [pdf])
Performance Evaluation, Volume 67, Issue 11, pp. 1155-1171

  Decomposition Results for an M/M/k with Staggered Setup
Anshul Gandhi, Mor Harchol-Balter and Ivo Adan
MAMA 2010 [pdf]
Performance Evaluation Review, Volume 38, Issue 2, pp. 48-50

  Analysis of an M/M/k system with exponential setup times under staggered boot up
Anshul Gandhi, Mor Harchol-Balter and Ivo Adan
MCQT 2010

  Power Capping Via Forced Idleness
Anshul Gandhi, Mor Harchol-Balter, Rajarshi Das, Jeffrey Kephart and Charles Lefurgy
WEED 2009 [pdf]

  Optimal Power Allocation in Server Farms
Anshul Gandhi, Mor Harchol-Balter, Rajarshi Das and Charles Lefurgy
SIGMETRICS 2009 [pdf]    (extended version, Technical Report CMU-CS-09-113 [pdf])

  Inducing Optimal Scheduling with Selfish Users
Paul Enders, Anshul Gandhi, Varun Gupta, Laurens Debo, Mor Harchol-Balter and Alan Scheller-Wolf
CMU GSIA Working Paper #2008-E23 [pdf]

  Database Summarization and Publishing in Wireless Environments
Anshul Gandhi and R.K.Ghosh
ICDCN 2006 [pdf]

Patents:

[Top]

   Provisioning Data Center Resources (filed 4/12/2012)

   Dynamic Capacity Management of Multiple Parallel Connected Computing Resources (filed 3/30/2012)

   Power Budget Allocation in Multi-Processor Systems (filed 4/7/2011)

   Systems and Methods for Managing Power Consumption and Performance of a Processor (US 8,589,709)


Teaching:

[Top]

   Fall 2014: CSE 691 Energy Efficient Computing


Service:

[Top]

   Program Committee Member: CCWS 2014, Sigmetrics 2014, HotCloud 2014, ICPP 2014, Sigmetrics 2010 (Shadow PC)

   Referee: IEEE Transactions on Computers, IEEE Transactions on Cloud Computing, ACM/IEEE Transactions on Networking, Performance Evaluation, Annals of Operations Research, IEEE Transactions on Parallel and Distributed Systems, IEEE Transactions on Services Computing, Canadian Operations Research Journal, IEEE Transactions on Network and Service Management, Sustainable Computing

   External Reviewer: ISCA 2013, HotPower 2012

   Webmaster: Sigmetrics 2013