Roja Eswaran, PhD

Senior Software Engineer at GEICO

[LinkedIn] [GitHub] [Google Scholar] [Resume] [CV]
Image
Bear

Publications

  1. Roja Eswaran, Mingjie Yan, Kartik Gopalan Template-aware Live Migration of Virtual Machines, In The Eighth ACM/IEEE Symposium on Edge Computing (SEC ’23) EdgeComm. Delaware, December 2023 [paper] [bibtex]
  2. Roja Eswaran, Mingjie Yan, Kartik Gopalan Tackling Memory Footprint Expansion During Live Migration of Virtual Machines, IEEE/ACM international Symposium on Cluster, Cloud and Internet Computing (CCGrid). Philadelphia, May 2024 [paper] [bibtex]
  3. Roja Eswaran, Mingjie Yan, Kartik Gopalan Incorporating Memory Sharing-Awareness in Multi-VM Live Migration, IEEE/ACM international Symposium on Cluster, Cloud and Internet Computing (CCGrid). Philadelphia, May 2024 [paper] [bibtex]
  4. Roja Eswaran Sharing-Aware Live Migration of Virtual Machines, Doctoral Dissertation [paper] [bibtex]
  5. Roja Eswaran Reducing Hypervisor Overhead for Virutal Interrupt Delivery in KVM/ARM Platform, Submitted as a part of Research Proposal Exam (RPE) conducted by Binghamton University on Fall’20 [paper]

Opensource Contributions

  1. Integrated new QEMU CPU-Pinning feature into EVE [commit1]
  2. Reduced memory usage of Idle Kubvirt from 1.1G to 350MB [commit1], [commit2 ]
  3. Upgraded Xen-tool version to take from 4.18 to 4.19 for EVE to take advantage of QEMU version upgrade from 5.1 to 8.0.4 [commit]
  4. Optimized traditional memcpy by only copying the valid data instead of holes thereby reducing copy time by 10 times [repository]
  5. Accurate QEMU Timestamps to measure the total migration time and downtime for live migration [repository]
  6. Measured memory allocation overhead introduced by QEMU [repository]
  7. Userfaultfd-Aware-Mmap to have fine-grained control over processes’ memory regions [repository]
  8. Kernel-level Timer and Inter-processor interrupt latency tests to measure the physical and virtual interrupt latency [repository1], [repository2]