Nagendra Singh Tomar

CTO

Bengaluru, Karnataka, India27 yrs 11 mos experience
Most Likely To SwitchHighly Stable

Key Highlights

  • 25+ years of experience in enterprise software
  • Expert in designing large scale systems
  • Strong focus on reliability and performance
Stackforce AI infers this person is a Cloud Computing and Virtualization expert with extensive experience in enterprise software architecture.

Contact

Skills

Core Skills

StorageCloud ComputingDistributed SystemsVirtualizationNetworkingSystems ProgrammingSoftware DevelopmentSystems Engineering

Other Skills

Linux KernelGoBlobfuseNFSAzure BlobOpenstackCephKVMVMware Virtual VolumesStorage Virtual ApplianceESX storage stackVVolStorage OperationsLinuxNetwork Replication

About

Being passionate about "extreme quality (Reliability and Performance)", great user experience and easy maintainability, I can convert any problem statement into an optimal and practical design.25+ years of experience working on various Enterprise Software products with specialization in:Linux Kernel, Storage (Ceph, iSCSI, NFS, VMWare Virtual Volumes, SAN), Networking, Cloud Computing, Openstack, Large Scale Systems, Containers (LXC, Docker), Server Virtulization, Qemu/KVM, VMWare vSphere, GNU/Linux eco-system, Computer Architecture (NUMA), System Performance. With excellent knowledge of complete computer architecture and GNU/Linux systems (kernel and user-space) in particular I find myself at ease in designing large complex systems that can scale well and perform efficiently.

Experience

Microsoft

Principal Engineer

Nov 2017Present · 8 yrs 4 mos · Bangalore · Hybrid

  • Working on enhancing storage for AI workloads**
  • <<Details cannot be shared>>
  • Add Distributed Cache Support to Blobfuse**
  • Blobfuse is the official Microsoft FUSE driver for accessing Azure Blobs. Users typically use it on their machines (VM or Containers) to mount an Azure Storage Container, once mounted Azure Blobs can be accessed as files in the mounted FUSE filesystem.
  • The goal of this project was to add a distributed cache layer using local NVMe storage from multiple participating nodes that form a cluster. The cache can be accessed in two modes - seamless and explicit. In seamless mode user accesses the files using unqualified path names and data will be served from cache if present, else it's fetched from Azure Storage. In explicit mode user could specify whether they want to access the cached copy or Azure Storage by qualifying the path with fs=[azure|dcache] using the first virtual pathname component.
  • As the lead architect I did the design and also implemented the key parts. The implementation is in Go and can be found at https://github.com/Azure/azure-storage-fuse/tree/feature/dcache
  • NFSv3 front-end for Object Store**
  • This is the Blob NFS service in Azure (https://learn.microsoft.com/en-us/azure/storage/blobs/network-file-system-protocol-support). I was the lead architect for this service.
  • It enables users to mount an Azure Storage Container as an NFS (v3) share to get an (almost) infinitely scalable NFS volume. Files created on this filesystem are saved as Azure Blobs allowing them to be accessed using the Azure Blob REST APIs and other supported interfaces like SFTP, etc. This was the MPA (Multi Protocol Access) feature.
  • Supports all NFS operations, while the target workload was large sequential IOs.
StorageDistributed SystemsCloud ComputingLinux KernelNetworking

Snapdeal

Senior Technical Architect

Sep 2015Nov 2017 · 2 yrs 2 mos · Bengaluru Area, India

  • At Snapdeal we successfully created one of the largest Openstack based private cloud (100,000+ cores, 4PB SDS) using 100% Open Source technologies - with Openstack, Qemu, KVM (Linux Kernel), Ceph (Storage) being at the heart of it.
  • I was the Technical Architect for the whole project and as the Architect I'm proud that we took many right decisions which helped us move Dev-n-Test fully to our private cloud in a short span of 3 months, and then in another 3 months the Production cloud was ready. Currently the snapdeal.com production has been fully migrated to our private cloud which was designed with two basic mantras - Robustness and Performance.
  • http://engineering.snapdeal.com/snapdeal-launches-its-own-cloud-snapdeal-cirrus-201609/
  • http://engineering.snapdeal.com/hybrid-cloud-a-path-between-the-seas-201609/
OpenstackCephKVMLinux KernelCloud ComputingDistributed Systems

Vmware

Staff Engineer 2

Sep 2008Sep 2015 · 7 yrs · Bengaluru Area, India

  • As one of the senior most Technical R&D members of the team, I was responsible for kick-starting lots of projects in VMware India. Along with leading various projects, I mentored junior team members and helped them grow. Some of the key projects that I was part of.
  • > VMWare Virtual Volumes (VVol)
  • http://www.vmware.com/in/products/vsphere/virtual-volumes.html
  • VMWare Virtual Volumes technology allows storage array vendors to extend their value added services to VMs. f.e. VVol enables compliant storage arrays to perform various storage operations like snapshot, clone, replication, etc at VM granularity.
  • I was the Architect for the VVol project and here are some of the note-worthy achievements.
  • Clearly defined the VVol API spec enabling consistent implementations from various storage partners with very few iterations.
  • Robust and performant VVol implementation for ESX, that storage partners could test against.
  • Helped storage partners extend VVol support to their arrays.
  • Numerous scale challenges.
  • > VMWare SVA (Storage Virtual Appliance)
  • http://www.vmware.com/in/products/vsphere-storage-appliance.html
  • SVA runs as a Linux virtual appliance that uses Linux MD-Raid1 for replicating across the network.
  • Lot of performance optimizations, including a novel read balancing algorithm that exploited disk layout to get 10-15% better random IOs/sec.
  • > Worked on various layers of ESX storage stack.
VMware Virtual VolumesStorage Virtual ApplianceESX storage stackStorageVirtualization

Adaptec india pvt. ltd

Technical Team Lead

Nov 2002Sep 2008 · 5 yrs 10 mos · Bengaluru Area, India

  • TCP offload engines (TOE) were becoming a craze those days. I was fortunate to be involved in one of the first TOE implementations (http://download.adaptec.com/pdfs/installation_guides/ana7711_v10_gs.pdf). This was implemented completely as a Linux kernel module which could add TOE support to a stock Linux kernel by live patching of kernel socket hooks.
  • Later, perceived, designed and implemented a generic TCP offload framework for Linux (2.4 and 2.6) with the intent of proposing to the Linux community for inclusion in Linux.
  • Various other interesting projects:
  • Porting Linux to custom ARM and PowerPC boards
  • Added zero copy networking receive support to get reasonable performance on these relatively "weak" CPUs
  • Misc performance optimizations like use of sendfile(), epoll(), batching SCSI commands, cache profiling, etc
  • iSCSI initiator
  • A Storage Virtual Appliance using Linux DM (device mapper)
TCP Offload EnginesLinux KerneliSCSINetworkingSystems Programming

Netbrahma technologies (networking startup)

Senior Software Engineer

Apr 2000Nov 2002 · 2 yrs 7 mos · Bengaluru Area, India

  • Worked on various modules of a Terabit Aggregation Router - IPC (Inter Processor Communication) for a Ethernet based fabric, IP forwarding plane, various SNMP MIBs, and other misc stuff.
  • Created one of the first working MPLS (Multi Protocol Label Switching) stack while it was still in IETF draft stage. The forwarding plane was implemented in Linux kernel 2.4.
MPLSIP ForwardingSNMPNetworkingSystems Programming

Texas instruments

Software Engineer

Feb 1999Apr 2000 · 1 yr 2 mos · Bengaluru Area, India

  • Developed a software solution for optimal placement of components in a chip, resulting in increased die efficiency.
  • This solution was used by TI's internal Chip Placement and Routing team and it increased the automatic coverage from ~30-40% to ~75-85%. It replaced a commercial EDA tool that the team was previously using.
  • All in all a sweet success and a huge moral booster!
Chip PlacementEDA ToolsSoftware DevelopmentSystems Engineering

Abb

Management Trainee

Sep 1996Jun 1997 · 9 mos · Bengaluru Area, India

  • Still an electronics enthusiast at heart who used to hate computers in school, this was my first job where I was introduced to the beautiful world of computers!
  • Mainly worked on ABB's industrial automation cards. I would say 80% h/w 20% s/w.
  • Most interesting and memorable part was solving a long pending problem in one of the CPU cards which was causing the bootup to fail. There were ~50 CPUs cards worth tens of crores lying around in the inventory, rendered unusable because of this bug. It was a multi-layer PCB with close to 300 major components.

Education

Indian Institute of Science (IISc)

MS — Computers

Jan 1997Jan 1999

HBTI

BE — Electronics and Communication

Jan 1992Jan 1996

Stackforce found 100+ more professionals with Storage & Cloud Computing

Explore similar profiles based on matching skills and experience