About
Cheng Huang is a research scientist and tech lead at the Microsoft Azure Storage team (opens in new tab). Lately, he has been leading the design, architecture and implementation of Azure Storage vNext (opens in new tab). He is also responsible for all erasure coding strategies in Azure Storage. Along the way, he has been introducing formal methods and model checking to the Azure teams.
- Leads the production development of Azure Storage vNext (opens in new tab) (to scale Azure Storage from EB to ZB).
- Responsible for all erasure coding strategies (within DC, within region and cross-region (opens in new tab)) in Azure Storage.
- Educates Azure teams on formal methods (such as TLA+ (opens in new tab) and IronFleet (opens in new tab)) and model checking (such as P# (opens in new tab)). Organizes the Dr. TLA+ Series (opens in new tab).
Dr. Huang has also been working on erasure coding technologies for storage applications. While the industry has been relying on conventional schemes, invented more than 50 years ago and primed in deep space communication, Dr. Huang and his MSR colleagues have partnered with the Windows Azure Storage team and designed novel schemes that are optimized specifically for cloud storage. The LRC paper received a Best Paper Award at USENIX ATC 2012. LRC significantly reduces storage cost over conventional schemes, while still maintaining required durability and I/O performance. Given the scale and growth of Azure, the savings from LRC translate into 100s of millions of dollars for Microsoft. This work has been covered by the Microsoft Research featured story.
In the past, Dr. Huang and his MSR colleagues have partnered with the MSN team, designed and developed Microsoft’s peer-assisted Video-on-Demand system, which significantly reduces the cost of video delivery while dramatically boosting video quality at the same time. Please check out the profitable VoD paper at ACM SIGCOMM 2007. This work has also been featured by New Scientist (opens in new tab).
Dr. Huang has partnered with Azure and built a Global Traffic Management system; the Apollo paper was presented at USENIX ATC 2010. Dr. Huang has partnered with the Xbox team, designed and built the core decision engine in Pangolin—the low delay network transport protocol used in Xbox social gaming—and presented the Pangolin paper at ACM CoNEXT in 2011. Dr. Huang has also partnered with the Bing team and investigated Internet Search hijacking. This work was presented in the Revealer paper at USENIX LEET 2011 and featured in stories by the Register (opens in new tab) and Ars Technica (opens in new tab).
Dr. Huang received his B.S. and M.S. degrees from Shanghai Jiao Tong University (opens in new tab), both in Electronic Engineering, and his Ph.D. degree in Computer Science from Washington University in St. Louis (opens in new tab).
[06.2016] Launching Dr. TLA+ Series (opens in new tab).
[08.2015] Teaching Tutorial on Cloud Storage (opens in new tab) at ACM SIGCOMM 2015.
[03.2015] Serving on USENIX FAST 2016 (opens in new tab)
[06.2014] IEEE Communications Society and Information Theory Society Joint Paper Award (opens in new tab).
[12.2013] Microsoft TCN Storage Technical Achievement Award for “outstanding achievement and contribution in Microsoft storage technology”.
[09.2013] LRC Erasure Coding shipped in Windows Server 2012 R2 and Windows 8.1. Here is my talk (opens in new tab) at SDC 2013 (opens in new tab) (PPT).
[02.2013] Jim Plank (opens in new tab) and I taught a Tutorial on Erasure Coding for Storage Applications (opens in new tab) at FAST 2013 (opens in new tab). Here are the tutorial materials (opens in new tab).
[09.2012] Gave a talk on “Erasure Coding in Windows Azure Storage” at SDC 2012 (opens in new tab). Here is the StorageMojo take (opens in new tab).
[06.2012] “Erasure Coding in Windows Azure Storage” (opens in new tab) receives a Best Paper Award at USENIX ATC 2012.
There is much renewed interest about Pyramid Codes. Here is an introduction.