Cupertino, CA / C++ / Linux / cloud storage
SivaprakasamSuresh
Performance & Distributed Systems Engineer
I build and tune high-throughput storage systems, networked data planes, and Linux/C++ infrastructure where tail latency, flow control, and small implementation choices matter.
+ About Me
Staff/Principal-level distributed systems and performance engineer with deep experience across distributed file, block, and object storage. I work close to the data path: fibers, AIO, epoll, OpenSSL, NUMA, NIC queues, contention, backpressure, and the measurements that separate guesses from fixes.
+ Writing
Beating Claude Opus 4.5 on Anthropic's Performance Challenge — 1,352 vs 1,363 Cycles
A technical performance write-up on cycle-level optimization and benchmark-driven tuning.
+ Systems I Build
OCI Extent Server messaging data plane
Tech lead for the messaging stack underneath Universal Storage Pools, powering object, block, and file services at cloud scale.
mTLS at high connection counts
Designed end-to-end mTLS for control and data paths with shared TLS context, reusable BIOs, zero hot-path allocations, zero extra data copies, and record batching.
Nutanix storage performance and scale
Led tuning across NFS, SMB, ZFS, distributed telemetry, metadata scale, caching, and AI/ML training data paths.
+ Performance Wins
Random writes +130%. Traced and tuned Nutanix filesystem behavior across the storage pipeline.
SMB writes +82%. Found a GRO interaction and turned a subtle network path issue into a measurable production win.
Shares scaled from ~300 to ~5,000. Compressed metadata, redesigned the data model, and accelerated control-plane queries.
SQL-like query layer over Cassandra. Built grouping, rollups, joins, downsampling, and always-on profiling for distributed telemetry/configuration stores.
+ Technical Depth
+ Experience
Oracle Cloud Infrastructure, Cloud Storage
Tech lead for Extent Server's messaging data plane.
Nutanix, Distributed Storage
Led performance and scale work across filesystems, query execution, observability, and AI/ML storage paths.
Juniper Networks, Routing/Switching Platforms
Led QoS features across classifiers, policers, shapers, RED/WRED, and packet-processing pipelines.
Earlier Roles
Worked on packet-forwarding infrastructure, network processors, optical switching, VoIP, IP fast switching, UFS, and NFS performance.
+ Patents
7 issued U.S. patents across distributed control-plane databases, high availability, virtualized-storage QoS, and OS/kernel memory internals.
+ Education
M.S. Computer Science, State University of New York at Buffalo
B.E. Computer Science, National Institute of Technology, Tiruchirappalli, India