Dahlia Malkhi is a Principal Researcher with the newly formed VMware Research Group. She is an applied and foundational researcher in broad aspects of reliability and security in distributed systems. She joined VMware Research as a founding member in December 2014. Prior to that she worked at Microsoft Research and the Hebrew University of Jerusalem. She holds a PhD, an M.Sc. and a B.Sc. in computer science from the Hebrew University of Jerusalem. Presently, she is passionate about the CorfuDB project she co-founded, a cloud scale consistency platform. CorfuDB open source is on github:corfudb.
Today, I am happy to share my excitement to be a founding member of VMware Research, established in December 2014, and residing in our CTO office in Palo Alto. Our small team includes Ittai Abraham, Marcos Aguilera, Dahlia Malkhi, Chris Rossbach, and Udi Wieder. We report to David Tennenhouse, VMware’s Chief Research Officer, and work closely with Curt Kolovson, Senior Staff Research Scientist for the VMware Academic Program (‘VMAP’). We have a full internship program lined up for 2015 with six interns on board.
We call ourselves the “VMware Research Group”, or “VRG” for short. Our motto is: Bring production to the VeRGe of research and research to the VeRGe of production; I will say more about this below.
By establishing a dedicated research group, VMware reinforces its long tradition of research focus and academic reach. It continues a history of cutting-edge technology innovation and a culture of technical leadership and market disruption. VMware was founded on a technology disruption grounded upon software-based OS virtualization. The company continued to surprise the market with ‘impossible’ breakthroughs like VM migration, VM forking, transparent page deduplication, memory balloons, and more. VRG-ers are proud to have joined VMware and to have become part of this culture of excellence, technology focus, and advanced development.
Here is a little teaser. Question: What’s in this image?
Answer: This is the first successful xerographic image, taken on the date shown by Chester Wilson, inventor of xerography.
Tremendous effort was required to turn this blurry page and the initially crude production process into the first commercially successful copying machine, the Xerox 914, which came out 21 years later! Then, almost overnight, it became a huge commercial success and turned into a billion dollar business within less than a decade. Later, to prepare itself for a future of digital documents rather than paper, Xerox founded the Palo Alto Research Lab (PARC), right around the corner from VMware’s headquarters in Palo Alto. PARC gave birth to the first laser printer, the first personal computer (Alto), the SmallTalk programming language, and half a dozen other “firsts”.
Although today’s software world is rolling out new technologies far more rapidly than the copying machine story, it is a still a fundamental truth that driving an invention to production takes a tremendous amount of innovation.
I firmly believe in the role that research plays in driving and accompanying the innovation efforts within a company. Research experts bring immediate value by providing in-house foundational knowledge and deep competencies in the topics the company cares about. Researchers also maintain external relations with the science world, help promote the company brand, and attract top talent. Over a longer horizon, research generates ideas that may push the envelope of possibilities. This has happened at industrial research labs in the past, and we are confident we can create the right framework for first-rate systems research at VMware. We build upon the experience of the best industrial research labs, as manifested in A Perspective on Computing Research Management, by Roy Levin.
To succeed, a research organization relies on operating in a two-prong manner:
- It must recruit leading academicians in relevant fields, and allow them to operate independently, while continuing to lead and impact the computing field. Research should not be siloed within existing development areas or organizational structure, because something special happens when a group of researchers are allowed to collaborate in a cross-domain, unfettered manner. It is greater than the sum of their parts, and it is what maintains a special and a unique edge.
- At the same time, industrial research is different from academia. It needs to align with the company mission and strive to bring real impact. It has to interact broadly across business-unit boundaries, listen to and understand the real challenges engineers face.
When touching and feeding each other, the deep-expertise prong and the broad-reaching prong establish a T shape:
This requires researchers to activity seek out core challenges and fundamental hurdles that developers encounter. The researchers find opportunities to bring confidence, abstract insight and foundational know-how. They proactively evangelize and “get on the roadmap” to drive impact. They create fruitful collaborations that feed back in a circular manner, feeding research excellence into production and bringing timely technology trends to research. They bring unique advantages, a deep and foundational look at problems, and push the envelope of what is possible.
Importantly, no framework could guarantee or force success, and at the end of the day, a lot depends on people’s readiness and agility to pursue and hold thorough and meaningful collaborations. VRG aims to build such a culture of deep and meaning relationships with the company by persistently maintaining a culture of deep interaction and collaboration, and by recruiting first-rate researchers who are keen to drive impact.
This timing is an exciting and transformative moment for a research organization to be established within VMware. VMware’s mission to empower information technology with efficient resource utilization and effective management tools is shifting. It needs to expand from running clusters of single-host software to a full-fledged data-center OS. It needs to incorporate federated and hybrid clusters running scale-out, distributed software services which interact across node boundaries, across service boundaries, and across administrative domains. VMware has a key position in transforming the shape of IT infrastructure.
VRG has already engaged in a number of exciting activities across a variety of domains. In the six months since our establishment, we have learned about various software efforts within VMware that have strong requirements for consistency, reliability and algorithmic foundations. We are building CorfuDB, an open-source transactional data platform built over a shared log. We are architecting Trillium, a virtualization layer for GPGPUs that implements efficient interposition. We are designing a service store for cloud providers. We are developing efficient copy-on-write index data-structures. We are working on geo-distributed transactions. We are working on log-analytics and sampling. And more.
VRG members actively collaborate with a large variety of research institutions around the globe. Our early 2015 publications include:
- Byzantine Agreement with Optimal Early Stopping, Optimal Resilience and Polynomial Complexity. By Ittai Abraham (VMware Research) and Danny Dolev (Hebrew University). STOC 2015.
- Low-distortion Inference of Latent Similarities from a Multiplex Social Network. By Ittai Abraham (VMware Research), Shiri Chechik (Tel Aviv University), David Kempe (University of Southern California) and Aleksandrs Slivkins (Microsoft Research). SICOMP 2015.
- Taming uncertainty in distributed systems with help from the network. By Joshua B. Leners, Trinabh Gupta (The University of Texas Austin and New York University), Marcos K. Aguilera (VMware Research), and Michael Walfish (New York University). EuroSYS 2015.
- Distributed Resource Discovery in Sub-Logarithmic Time. By Bernhard Haeupler (Carnegie Mellon University) and Dahlia Malkhi (VMWare Research). ACM PODC 2015.
- SurroundWeb: Mitigating Privacy Concerns in a 3D Web Browser. By David Molnar and Eyal Ofek (Microsoft Research), Chris Rossbach (VMware Research), Alex Moshchuk (Google), Helen Wang and Ran Gal (MSR), John Vilk (University of Masschusets), Ben Livshits (MSR) . IEEE Security & Privacy 2015.
- Albatross: Systems Support for Augmented Reality. By Christopher J. Rossbach (VMWare Research) and Emmett Witchel (University of Texas at Austin). SFMA 2015.
- REEF: Retainable Evaluator Execution Framework. By Markus Weimer and Yingda Chen (Microsoft), Byung-Gon Chun (Seoul National University), Tyson Condie (University of California Los Angeles), Carlo Curino and Chris Douglas (MS), Yunseong Lee (Seuul), Tony Majestro (MS), Dahlia Malkhi (VMware Research), Sergiy Matusevych (MS), Brandon Myers (University of Washington), Shravan Narayanamurthy, Raghu Ramakrishnan, and Sriram Rao (MS), Russell Sears (Pure Storage), Beysim Sezgin and Julia Wang (MS). SIGMOD 2015.
- Approximate Nearest Neighbor Search in Metrics of Planar Graphs. By Ittai Abraham, Shiri Chechik, Robert Krauthgamer, Udi Wieder. Random-Approx 2015
Check us out at http://research.vmware.com and look for more updates from VRG to come!