Cbench
Automates the HPC benchmark lifecycle—from generating scheduler jobs to parsing output into structured results.
31 benchmarks · Perl + PythonChris Maestas · CTO, Data & AI Storage Solutions — IBM
Data infrastructure for HPC and AI.
I design systems that put unstructured data to work—from edge clusters to global-scale storage—and build open-source tools for benchmarking, cluster operations, and IBM Storage Scale.
Selected work
Tools built for real infrastructure.
Personal open-source work focused on repeatability, safer operations, and useful data at cluster scale.
Scale GUInstall
A guided web interface for installing, deploying, and upgrading Storage Scale clusters, with command previews before execution.
Signed RPM + DEB packagesNetwork Readiness (KOET)
Validates latency, throughput, and packet loss before a Storage Scale deployment goes near production. Forked from IBM/SpectrumScale_NETWORK_READINESS.
Preflight for cluster networks · ForkMore projects
Cluster and data tooling.
CAP — Cluster Administration Package
Integrates common HPC operations tools behind a unified build and install system with RPM and Debian packaging.
Storage Scale MCP Server
Exposes IBM Storage Scale REST APIs and CLIs as structured tools that language models can use with a live cluster. Forked from IBM/ibm-storage-scale-mcp-server.
ESS Comprestimator
Samples existing data to project how much capacity compression could reclaim before enabling it. Forked from IBM/ess_comprestimator.
dool
Real-time command-line monitoring of Linux system resources—CPU, memory, network, disk, load—with a plugin architecture. Forked from scottchiefbaker/dool.
Storage Scale Agents
Reference AI agents built on the Storage Scale MCP server, giving language models task-focused ways to operate a live cluster. Forked from IBM/ibm-storage-scale-agents.
GPFS User Group Tools
Community tools from the Spectrum Scale (GPFS) User Group for administering and inspecting Storage Scale clusters. Forked from gpfsug/gpfsug-tools.
pdsh
A multithreaded parallel remote shell client that runs commands across many hosts at once, over ssh, rsh, or Kerberos IV. Forked from chaos/pdsh.
ECE Capacity Estimator
Calculates effective usable capacity for Storage Scale Native RAID (ECE) systems from a hardware configuration, or solves in reverse for the smallest configuration that meets a capacity target. Forked from IBM/SpectrumScale_ECE_CAPACITY_ESTIMATOR.
ctrl-data-mgmt
A local file metadata catalog: index the directories you care about, then find what's eating your disk and what you're storing twice. Pure Python standard library, nothing leaves the machine.
Ideas in the open
Publications, talks & writing.
Selected work on storage architecture, AI data infrastructure, and operating systems at scale.
IBM Redbooks
Talks & media
About
Systems at scale.
For more than 25 years, I’ve architected enterprise IT systems across federal, commercial, and public sectors. My work spans HPC, hybrid and AI infrastructure, and unstructured data across file, object, and tape.
I work across cloud, on-premises, and edge environments on bare-metal, virtualized, and containerized platforms. The projects on this site are personal open-source work.