Larry Clapp Email & Phone Number

Walland, Tn, Us

Oakland, Ca, Us

Solopreneur & bootstrapper.hucksh(ell) is a new shell for Unix-like systems that combines elements of bash, screen/tmux, ssh, sqlite, and more. Sign up for the mailing list at https://mailchi.mp/e71e8020b1b5/sign-up-for-news-about-hucksh.Hucksh is written in Go (Golang), and uses Gio UI, gRPC, SQL, and SQLite. It is available for macOS and Linux, and a client-only version for Windows.

San Francisco, Ca, Us

Worked on the Kentik back-end. Kentik processes billions of records across hundreds of servers, with near-realtime querying.Most of the back-end is in Go (golang); all the below work was in Go. Other technologies used: Github, Linux, gRPC, tcp & udp networking, Puppet, Grafana, Kibana, Jenkins, Envoy, Lightstep, Nomad.Working at Kentik was fun and challenging. Like many SaaS applications, you had to deploy fixes and upgrades *while the application was running*.Highlights:* Built a custom load balancer that dynamically reallocates flow processing across 70+ hosts based on various constraints such as "flows per second" and cpu/memory usage. This allows Kentik to use its backend more efficiently and helps reduce CapEx, and also makes the back-end more resilient to internal outages, by quickly detecting when a host goes down and redirecting the data that had been processed there.* Modified almost every part of the backend to allow processing multiple client devices in a single process. This required changes in the ingest, storage, and query layers. As a direct result of this facility, Kentik was able to sign several large "household-name" clients that it couldn't have handled previously.* Modified the ingest part of the back-end to restart more efficiently and drop less data, e.g. during system upgrades.* Wrote the Kentik Streaming Telemetry (ST) sub-service; ST replaces and is more efficient than SNMP. ST is frequently implemented via gNMI (gRPC Network Management Interface). It has a "standard" but writing code against any given manufacturer is still challenging. I wrote "dial-in" ST, where Kentik "dials in" to client machines, for three different major switch manufacturers.* Participated in Kentik's "Ingest 2.0" design discussion, where most of the backend team worked up a new way to process data, using Kafka, Apache Arrow, and Apache Parquet.

Lisbon, Lisbon, Pt

Refactored the Call Correlation proof of concept, a batch-oriented approach using Hadoop, to a production-ready streaming approach using WAF, Perl, and Redis; did some preliminary work in Google Go. So far as we know, no one else in the world does call correlation or even knows how to do it, including any of our clients or their switch providers.Designed and implemented Rural Call Completion, a fork of the netCLARUS backend.Refactored and rearchitected cdrWatch into netCLARUS, able to process Call Data Records (CDRs) from a “Tier 1” client (7-8 billion CDRs a day), on a Condor cluster of several CentOS Linux servers. netCLARUS is a Perl & MySQL data processing application talking to a .Net & MS SQL Server front end.Stress-tested the Condor cluster to prepare for our first Tier 1 client, ironing out problems with the hardware and software ahead of time, before the client was actually signed. Built the cluster so that servers can be added with a minimum of fuss.Managed and helped develop all later versions of netCLARUS. Added new switch types and new reports. Continued refactoring to increase code quality, readability, reliability, and performance.Trained new team members in netCLARUS internals, programming in general, OOP, Perl, Go, Vim, Scrum, TDD, and other team practices and processes.Assorted other development, scripting, and system administration, typically using Perl, bash, and zsh. Managed my own Debian Linux virtual machine using VMWare, and then Oracle VirtualBox. Some work in Google Go.Worked remotely, from home, 80% of the time.Note that this includes my time with Connectiv (before WeDo purchased it), as well.

Bethesda, Md, Us

Architect and implement the next version of routeWatch & cdrWatch. Redesign cdrWatch into netClarus. Scale netClarus to handle "Tier-1" clients, billions of CDRs a day.Connectiv merged with Wedo Technologies in May 2012. See my Wedo entry for more experience.

Hebron, Ky, Us

IBM Global Services, Tivoli Migration & Support.With another team member, architected and implemented Tivoli monitoring hierarchy for 3500 servers.Designed and implemented T/EC "fact file" generation and T/EC event routingHeaded rollout of full disk monitoring throughout DisneyDesigned and implemented Tivoli agent and monitoring engine monitor. Automated automatic fixer script; reduced on-call paging by at least 50%.Developed and documented several procedures to streamline team use of client tools (Service Center, Remote Desktop, etc).Saved the Openview team several person-months of work by automating parsing of Openview monitoring templatesWorked remotely, from home, 100% of the time.Worked with AIX 5.3, Red Hat Linux, Windows XP, Tivoli Monitoring (ITM 5.1), Tivoli Inventory (ITCM 4.1), Tivoli tasks, Perl, ksh, zsh, Perl "yacc", Service Center

Whitehouse Station, Us

Helped design and write VzWorks, a web-based, trackable, auditable form-based instant messaging client written in Java and JavaScript, using an XML message format, and IBM/Lotus SameTime for message delivery. VzWorks eliminates the communication overhead between front-line customer service representatives and rear-line repair dispatchers, while prohibiting an interaction from degenerating into non-business-related chatter.Developed a configurable metrics generator for VzWorks in Java, which automatically creates & populates tables, and generates XML reports, showing specific performance and system metrics for the VzWorks system.Learned, advocated, implemented, and taught Test Driven Development and Automated Unit Testing.Worked with AIX 4.3/5.1, DB2, Informix, SQL, Continuus, JSPs, XML, XSL, and Debian GNU/Linux

Us

Tivoli maintenance and development. Worked with Chris Cox (among others).Automated Tivoli gateway assignment via Perl scripting of Endpoint Policy scriptsDocumented and debugged Tivoli Server Performance Prediction toolsDeveloped and documented an automated Tivoli Endpoint upgrading system in Perl.

Bethesda, Md, Us

Tivoli endpoint deployment, maintenance, and developmentLead engineer in successful deployment of Tivoli's Remote Control product to 20,000 users across the USDeveloped instructional material and taught class on Remote Control to Help Desk personnel. Developed an endpoint tracking system in Perl. This system included an endpoint database, maintenance utilities, and associated processes to keep the data in the Tivoli system accurate.Worked with Tom Cooper (among others).

Maintained and extended Oasys, a large, multi-user, ObjectOriented (C++) Trade Capture/Management systemOn my own initiative, fixed window resizing for several commonly used windows in Oasys. Many users commented that that was their favorite feature of that release.Designed and wrote the Mass Order Generator prototype in TaskMate, a proprietary 3270 scripting languageWas pivotal in the MOG requirements gathering meetings; received many compliments from meeting participantsOperated well autonomously with little managerial input or supervisionDirectly supported a 10 person reporting team using the BusinessObjects report writer and a Sybase database systemTrained users on BusinessObjects and Paradox

Wrote the Automated Testing System, a multi-user program to automate program testing. Seven years later, in May 1999, ATS was still in use at GTEDS, virtually unchanged.Wrote a code generator for Embedded SQL in awkHelped design and implement several new revisions of CMSS, a multi-tier, distributed, client/server order collection toolSupervised a 4 person development team and an 8 person maintenance teamDeveloped many shell/awk/Perl scripts; functionality included report generation, data gathering, system administration, etc.