Manager Service

Punya Bheemesh Akula

CONTACT Email: ***********@*****.***

Phone: +91-913*******

PROFILE SUMMARY

WORK EXPERIENCE

Have a total work experience of 7 years in Linux Kernel Programing & Linux Internals & System Programming, Networking, C, C++, Java

Extensive experience in Linux core operating system internals and system programming

Good working experience in C, C++, java programming languages

Working experience in Network drivers, Network processor’s and kernel programming

Extensive understanding of Linux Network Stack (L2, L3, L4 layers) and Networking Protocols(TCP, UDP, ICMP, STP, ARP)

Working experience in developing OFED stack (Open fabric enterprise distribution) which based on

RDMA.

Working experience in DPDK and DPDK-OVS

Good knowledge on Algorithms and Data Structures

Good knowledge on PCIE technology.

Experience in contributing in various projects and has experience in Analysis, Designing, Developing and Testing.

Experience in debugging, troubleshooting on Customer network and handling critical customer cases

Passionate to work in Networking & Network Virtualization like SDN(Software defined networking)

Broadcom Corporation, Hyderabad, India - 500081 June 2014 Till Date

Staff II Engineer

DPDK Poll Mode Driver For XLP Network Processor: XLP is mips based network processor which has 16 cores and each core has 4 hyper threads (appears like 64 cores). Added poll mode driver for this NIC interface into dpdk, developed one kernel module to provided sufficient information need by the poll mode driver. Used NETLINKsocket for the communication between poll mode driver in user space and kernel module. Added feature of distributing the traffic among queues based on the destination mac address. Added feature of TSO into DPDK and modified the TESTPMD application to use both above features.

oContribution:

Developed poll mode driver as dynamic library

Destination MAC based routing of traffic into queues is added into DPDK

TSO support added into DPDK

Destination MAC and TSO commands are added into TEST PMD to test them

Network Driver for XLP processor : XLP is mips based processor which has many modules like NAE (Network acceleration engine), POE (packet order engine), SAE (security acceleration engines) and etc. Network driver interacts with the network acceleration engine and provide the Ethernet based communication. Driver used NAPI mechanism while receiving the packets, TCP segment offloading (TSO), L4 checksum, flow based routing using hardware parser, all these feature are notified by driver to network stack as hardware support them.

oContribution:

Added NAPI mechanism while reading packets

Registration with network stack and proper handling of TX and RX paths

Bug fixes in the some of the features

One Convergence Inc, Hyderabad, India – 500033 June 2011 --- May 2014

Software Engineer

Service VM Chaining and Life cycle Management: This is the process of integration of Network Service Virtualization with Openstack. Service VM Chaining is one of the two core parts of Group Based Policies(GBP). Service VM life cycle management is done by Service Manager and it involves right from the process of provision of the service VM to decommission of the same, which serve LBaaS, FWaaS, VPNaaS functionalities. Service Manager fits the service VM in Openstack environment and handles the requirements of each vendor according to vendor specifications and capabilities.

oContribution:

Added a new feature to Service Manager which is persistent VM parking

Developed a module in Service Manager to handle load balancers

Replaced default LBaaS agent with custom LBaaS agent to work with Service Manager

Added module to handle 3rd party load balancers

Capella 2 Network Switch ( RDMA based like mellanox): Capella 2 is like switch which uses PCIE concepts to drive traffic among end nodes. It is completely through DMA engine based. For this high speed switch needs compatible drivers at data link layer for end nodes. Design had two layers Link layer(LL) and Nic Layer(NL). NL works as intermediate layer between kernel TCP/IP stack and LL. Apart from this OFED layer(like NL) on top of LL also developed to get zero copy(which is not possible in TCP/IP) right from user space to kernel space, to get kernel by pass while driving data(this is most advanced way of doing networking).

oContribution:

Developed Ethernet layer which registers with Linux network stack as NIC interface

Developed OFED layer which registers with Linux OFED stack and uses infiniband network communication instead of Ethernet

OFED Stack for PLX: They are user space libraries which communicate with the kernel module and providing the basic environment for the TX and RX without any further intervention of kernel module. The memory used for TX and RX is pinned and shared between the user space and kernel space so that packets queued from users pace and directly hardware can use them via physical address space, this way avoiding the kernel copy while transferring and receiving the packet.

oContribution:

Developed user space libraries and kernel space driver for environment setup

TX/RX paths in the kernel driver using queues distribution and NAPI kind of mechanism while handling interrupts.

Entire OFED stack development for PLX RDMA based device.

Motorola, Hyderabad, India – 500081 June 2010 – April 2011

Intern

Moto Agent: Agents are like access points for the customer premise devices, these are communicated among them to provide the clustered network. These points are managed and configured by the management system which is one configure these agents to work as network. In real work they are like cell towers working in coordination mode and providing the customer disrupted service.

oContribution:

Completed developed in the C++, added couple of bug fixes

Management sub system is developed in JAVA which communicated with agents, added couple of bug fixes.

Thank you for using www.freepdfconvert.com service!

Only two pages are converted. Please Sign Up to convert all pages. https://www.freepdfconvert.com/membership

Contact this candidate