sumit gaur

Hewlett Packard, Bangalore (13 September 2010 till date)
Presently working with HP, Bangalore as Senior Specialist for HP CloudSystem Matrix. HP CloudSystem
Matrix is Infrastructure-as-a-Service (IaaS) for private and hybrid cloud environments, allowing users to provision infrastructure in minutes for physical and virtual. CSM also provide means to migrate, provision,
monitor and configure virtual guests. CSM supported hypervisors include VMware, Microsoft® Hyper-V, and Integrity VMs.
I am working for Integrity stack components of CSM. I am also being involved in completion of designing and development of different use case for HP Private Cloud strategy.

1. HP Integrity Private CloudSystem Matrix
HP CloudSystem Matrix combines automated design and provisioning of physical, virtual machine and applications through a self-service portal-with capacity planning and disaster recovery.
I am part of the team to analysis and decide the future road map items to complete the HP Private
CloudSystem matrix for Integrity servers. This exercise includes preparing whitepapers, designing the architecture diagrams, defining the interfaces and doing an impact analysis of new features to complete the
private cloud use cases for integrity servers.
My current assignments are dynamic configuration management and real time alert mechanism.

2. HP Integrity Virtual Machines Management Provider
Role and responsibility: Responsibilities include leading the designing, development and testing virtualization provider.
Contribute to plug-in integration strategy for different framework components to provider and hpuxpitool.
Time line: going on

Project Module: Virtualization CIM Provider
Environment: C++/WBEM/CIM on HP-UX.
Project Detail: The HPVM VirtProvider project had designed and developed a CIM provider that is conformant to the

Sun MicroSystems, Bangalore (16 May 2005 till 9 September 2010)
I had worked with Sun Microsystems as MTS. I had been involved with designing, development, and debugging of Magnum, Netra and Sun Cluster management software. During this period I had done
following major project:
1. Sun fabric management software for Constellation System (Sun's HPC initiative)
(Sun Constellation system was No. 6 system among world's TOP500 Supercomputers, achieved 433.2
Tflop/s, now project is part of Oracle Exadata system)

Role and responsibility: Magnum is the code name for Constellation system that will be used to implement complete HPC
cluster solutions from Sun for data-centers based on Infiniband. I am working on different modules of fabric
management software for Magnum. I was in charge for any new and on-going design/development/test in project at india center. I was also responsible of the delivery of the releases and of the technical
documentation. I have contributed in the following modules of project.
Time line: approx 3.5 years.

Project Module: Fabric Director (FD)
Environment: C on Linux/Solaris using Oracle Berkeley DB apis, SUN RPC, mxml lib and Posix threads.
Project Detail: Fabric Director aggregate physical view of the complete cluster fabric in terms of physical enclosures,
FRUs, connectors and related cabling, as well as the logical view of the fabric as a collection of switching
elements and end-nodes with associated attributes and connectivity.
The FD Core Services provide a multi-session client interface which supports both synchronous
queries and control commands from the clients to the FD as well as asynchronous events from the FD to the clients. FD communicates with different components using methods include CLI, RPC, SNMP, PTMX.
FD correlates the connectivity and state information to validate rules of how the cluster should be
constructed. FD is also responsible to do config management, fault management and performance
management by auto discovery of cluster. The FD can assist in resolving connectivity problems and link
problems by identifying all involved components as well as the state of those components. The Fabric
Director will be extended to support ILOM interfaces that intern provides SNMP/IPMP apis for the information gathered at FD.

Project Module: Link Diagnosis Agent (LDA)
Environment: C on Linux/Solaris using SUN RPC, Posix pthreads and OFED apis by mellanox.
Project Detail: Sun IB management software has LDA in the agent layer. LDA collects inband link/port diagnostic
information for the local-Host/switch IB port connectivity to the magnum subnet. The information is
processed and reported to the upper layers as required.
* LDA provides generic link state and unique port-ID information via async events.
* LDA provides aggregated counter values as well as historical logs via queries (both generic and HW
specific counter types can be provided)
* LDA provides bandwidth utilization, degradedness and congestion for all the active links.
* LDA also supports administrative operations include initiation of special diagnostics operations as well as changing the set of monitored links and adjustment of monitoring parameters like sampling rates and
threshold values for various links states.

Project Module: Channel interface Provider (CIP)
Environment: C on Linux/Solaris using Mellanox OFED (OpenFabrics Enterprise Distribution) libraries.
Project Detail: CIP instance represents an abstraction of Infiniband node (switch, channel adapter or router) and part of agent tier.
* CIP instance is responsible for converting the header and payload information specified for outgoing
packets from LDA (Link Diagnosis Agent) into a format that allows the underlying HW to present a
corresponding IB compliant packet on the output port.
* CIP is responsible for converting the incoming packets from the format that underlying HW provides to the format used by LDA.
* The CIP instance must make sure that outgoing packets are routed according to the address information that
is associated with the packet.
* For incoming packets, the CIP instance must verify that packets are in accordance with the IB spec defined
acceptance requirements for the port and interface.
* Packet Flow Control
* Error handling

Project Module: Fabric Director Proxy (FDP)
Environment: C on Linux/Solaris using SUN RPC.
Project Detail: Sun IB management software has FDP in its middle tier for exporting information about the state of the fabric and its components. To make complete fabric management system a scalable solution we need to
have another module Fabric director proxy (FDP) between FD and LDA. Since the number of nodes in large
clusters can be several thousand, it is in general not possible for the FD to communicate directly with every
individual node. Intercommunication between FD, FDP and LDA would be on RPC based connections.

* The FD monitors information from a number of Fabric Director Proxies (FDPs)
* Each FDP instance is identified by an IP address that is associated with a Chassis/Shelf Manager, a
Server/Host Service Processor or an IB Host (Os) itself.
* After gathering all HCAs information of one subnet, FDP will create a consolidated list and send back to
FD for keep all records in database.

2. Sun Cluster: Role and responsibility: I have worked as individual contributor and was responsible for develop/test
complete module for both Linux and Solaris platform. My work in this project was to made complete
changes related to management console and clquery demon. Management console related changes include
support for managing and monitoring Europa nodes. These changes deals with JMX/JDMK technology, JSP,
Java scripting and xml technology. Whereas clquery demon which provides RPC interfaces for management
console to interact with other library. Clquery demon changes deals with Sun RPC technology and C++.
Timeline: 1 year

Environment: C/java on Solaris using JMX/JDMK technology.
Project Detail: Project deals with Sun Cluster horizontal enhancement from 16 nodes support to 1024 nodes support. Sun
Cluster is a multi-system, multi-site disaster recovery solution that manages the availability of applications
services and data across local, regional and vastly dispersed datacenters.

Reliance Infocomm Ltd., Mumbai (2 June 2003 till 15thMay 2005)
I have worked with Reliance Infocomm, Mumbai as a L3 Executive (Software Engineer) . Involved with the development, investigation, standardization, evaluation, adaptation and management of key technologies for
current and future information networks. During this period I have done following major activities:
1. Prepaid LCO: Role and responsibility: I have worked as individual contributor and was responsible for implement/test complete module and integrate it with Rating Engine and Service logic using SUN RPC interfaces. Implementation includes logic to follow different predefined policies to allocate talk time to a prepaid subscriber.
Environment: C++ on Solaris using SUN RPC.
Project Detail: This project has three major modules AMS (Account Management System), Rating Engine and Service
Logic. I was working on AMS module. This module was responsible for maintaining the subscription details
relevant to an account and the account state information of all the accounts. The interfaces for the AMS
should be made open and published for external system. Hence the external systems should be able use the services provided by AMS. The interfaces are developed using RPC technology. Complete project developed in C++ with complete SDLC.

2. RapLoadGen:
Role and responsibility: I have worked as individual contributor and was responsible for implement/test C++ based core module of fully automated application testing system. RapLoadGen automates the complete process from load
Generation to report preparation at the end of the test.
Environment: C++ on Solaris.