In the true spirit of serving the larger cause of securing applications, machines are employed to theorize, test and confirm the impact of any security event. The alert models should be trained to only notify the user if it detects any suspicious activity, attack or breach, based on continuous monitoring of every action of every cloud component at every layer all in near real time.

Unshackle from the old security paradigm of - perimeter, rules, policies. Surpassing reactive management using log parsers and analyzers. Its time to take the problem head-on and create disruptive technology using deep temporal baseline to track, adapt and model entities, behaviors, connections and associations... paint the full security picture.

Design and development of data plane in three generations of Nimble's data storage platform with key focus areas of -

* Storage Capacity

* Data Protection (Availability and Reliability)

* Significant improvement in performance

Design and development of Infosight connectors - cloud-based storage management and support service (management plane) within the data storage platform for -

* Proactive platform and IO Stack health monitoring

* Performance characteristics of Flash memory - WAF and IOPS metrics

* Proactive component modeling for correlative and accelerated failure using telemetry data

Publish Design Requirement Specifications -

* Guidelines for third party components used in data storage platform

* Performance Modeling (end to end)

* Component Fingerprinting and integration with telemetry data

Led cross functional initiatives in design & development of platform services, RAID, Block IO & optimization of Mediagrid Clustered Storage File System built on Linux

• Implemented extensive changes to Linux IO stack & IO scheduler

o Large block access

o Enhanced error handling in SAS HBA driver

o SCSI mid layer enhancement for performing analysis of command issuance & completion

• Design & development of efficient job control framework for self-contained tasklets

o Eliminating excessive movement of data across storage nodes saving cluster bandwidth

o Infrastructure used for media transcoding & simultaneous broadcast across different media taps

• Key contributor in evaluating security vulnerabilities & performance enhancements in the Linux kernel

• Wrote specification & made recommendations on guidelines for platform hardware/firmware enhancements to match design parameters of Mediagrid Clustered Storage File System

• Authored white paper on proactive disk drive failure detection based on continuous throughput & latency measurement. The technique uses disk error classification & rate of occurrence metric to manage disk health

• Design online data evacuation from storage node using block move constructs while the cluster is online

• Design & develop technique to provide bounded disk latency

o Optimally conditioning device latency & compute requirement

o Ensuring reliability & availability of Mediagrid Clustered Storage File System with minimal performance degradation when the cluster nodes are sub-optimal

• Design & develop background data/metadata scrubber

o Runnable during idle periods

o Used smart drive recovery – incremental recovery by reconstructing chunks within storage node (parity) or replicating bad chunks from other nodes

• Published requirement specification for 3rd party component integration of next generation Mediagrid platform. Collaborated with external engineering team in design review & final deliverables of software & firmware

Led cross functional, multi-site team in feasibility study and prototyping of a portable RAID layer capable of being run seamlessly on a HBA with a dedicated local processor or using the Host Processor. The prototype solution was developed using an Intel Zion local IO processor.

Led efforts in building an offshore engineering tiger team by creating TOI’s and a mentorship program for continued development, feature enhancements, and customer design engineering of Host based RAID products.

Designed, Developed, and Integrated SATA fan out Port Multiplier (using Silicon Image SI3726) to HBA’s that were running Host based RAID solution. Implemented queue management techniques to avoid IO resource starvation, device throttling, and support fault tolerance. The RAID core support is an in-band transactional layer within the SCSI driver model, operating on a family of Parallel SCSI and SAS HBA’s. Additional API were designed and developed for RAID management, running as agent services providing hardware independent abstraction. These agents export API to Adaptec’s proprietary GUI and have hooks for plugging industry standard SNMP and CSMI management interfaces.

Design study and development of iSCSI protocol (based on draft 10) using Policy Specific Engines (PSE) and TCP hardware offload engines. Collaborated with the CTO and lead hardware designers in mapping the vision and capability of the end-to-end solution. Worked with a team of focused people in building the iSCSI target interface layered on top of a generic SCSI Host bus adapter providing connectivity to the underlying storage fabric. The iSCSI specific details were partitioned between driver and firmware to make optimal use of resources and provide capability for fine-tuning the solution. A management interface was created to manage, monitor, and fine-tune performance policies of iSCSI HBA.