Henglin is a technologist with 15+ years of experience of building enterprise storage software who enjoys solving challenging problems, designing and implementing solutions. He likes running, and enjoys playing with his son.

Technical Leader II Engineering @ Redesigned and implemented snapshot feature in a kernel driver, improving write performance by more than 10 times with multiple snapshots in comparison to previous snapshot implementation, reducing performance impact of write by more than 20% with single snapshot. Designed and implemented storage virtualization replacing LVM layer. Designed and implemented storage thin provisioning. Designed and implemented APIs to get changed locations between two snapshots, getting rid of expensive checksum computation to get the delta, and reducing replication time by up to 50%. Designed solutions to handle concurrent I/O to active-active mirroring server nodes in a HA environment. Designed SCSI persistent reservation handling with virtual devices in HA environment. From April 2014 to Present (1 year 7 months) Whippany, NJMember of Technical Staff @ Changed the architecture to scale upon number of CPU cores. Improved IO performance from 75,000 IOPS to 495,000 IOPS per NSS node. Enabled further improvement for IO performance. From October 2013 to April 2014 (7 months) 2 Huntington Quadrangle, Suite 2S01, Melville, NY 11747Software Engineer - Technologist(II) @ Designed integration of VeloBit caching software with Virident Firmware From July 2013 to October 2013 (4 months) Lincoln, MassachusettsSenior Principal Software Engineer @ Design, implement and review non volatility. Porting caching to VMWare ESXi. From January 2012 to July 2013 (1 year 7 months) 55 Old Bedford Road, Suite 203, Lincoln, MA 01773Director of R&D @ Designed, developed and maintained IOCore the data path module of IPStor. Ported IOCore module into user space and greatly increased system stability. Designed and implemented storage cluster link updating the HA secondary server with most current Timemark and safe cache metadata, making HA failover time constant regardless of amount of metadata for both features. Optimized Timemark storage allocation and time to delete a huge Timemark becoming constant. Implemented processor device accepts a set of vendor specific commands to copy data from remote SIR servers greatly improving the SIR replication speed. Implemented Timeview replication to save the Timview changes to a remote replica server. Designed and implemented Thin provisioning Disk to use less physical storage and allocating physical space on demand. From February 2007 to January 2012 (5 years) 2 Huntington Quadrangle, Melville, NY 11747Software Engineer @ Implemented Continuous Data Protection to recover to any point of time of a protected storage disk. Optimized Timemark memory usage dramatically increasing number of created Timemarks. Designed and supervised the development of MTCP protocol with multiple TCP connections to transfer data between two replication servers. Optimized safe-cache metadata speeding up the loading process of safe-cache enabled device by more than 100 times. Designed and implemented adaptive replication continuously backing up the data to a remote site and ensuring minimum data loss when the primary storage goes offline. Timemark storage metadata integrity online checking and correcting utility. Callhome monitors the health of IPStor server and reports through email if it detects anything wrong. VBDI block device driver for Solaris to support NAS on Solaris platform. Later ported into Linux and used for both NAS and FDS. Advanced backup for IPStor enabled users of IPStor to backup the content of a virtual device. The backup can be full, incremental or differential. The full backup will keep a copy of the whole virtual device. The incremental backup only gets all the changes since last backup. The differential backup obtains the only changes since last full backup. Online information collecting enables IPStor to track MTU negotiation, packet retransmission, and each packet size for fixed amount of time for client/server connection. The information greatly sped up the troubleshooting and helped improve the performance of some non-ideal network environment to improve replication performance. Implemented wrapper functions of Linux kernel functions for Solaris making porting IPStor between Linux and Solaris effortless. A SCSI device driver to collect Linux SCSI device information to enable IPStor to simulate correct query information. Snapshot and data integrity testing utilities to help debug snapshot logic. From June 2001 to February 2007 (5 years 9 months) 2 Huntington Quadrangle, Melville, NY 11747Assistant Researcher @ Small Angle X-ray Scattering Data Analysis software development in Fortran Data acquisition system programming in C and Assembly From July 1995 to July 1997 (2 years 1 month) Beijing City, China

M.S., Computer Science @ Stony Brook University From 2000 to 2001 M.S., Physicc @ Institute of High Energy Physics From 1992 to 1995 B.S., Physics @ Tsinghua University From 1987 to 1992 Henglin Yang is skilled in: Storage, Storage Virtualization, Fibre Channel, Cloud Computing, Virtualization, SAN, Linux, iSCSI, SCSI, Storage Area Networks, High Availability, NAS, Kernel, File Systems, Linux Kernel, VMware, Enterprise Storage, Unix, Storage Architecture, VMware ESX, Solaris, Disaster Recovery, Servers, Storage Management, Replication, Storage Solutions, Data Center, Server Architecture, System Architecture, Cluster, Deduplication, Distributed Systems, Shell Scripting, Performance Tuning