|Name||P.B. van Pelt|
|Date of Birth||July 6th 1976 (Nuenen, NL)|
|Marital Status||Committed, one child (2007)|
EducationGrammar school at Sint Jacobs Lyceum, Boxtel, Netherlands (1989-1996), Technical University of Eindhoven (1996-1998), Fontys Technical College of Eindhoven (1998-2002), Bachelor of Science received cum laude.
Active board member at MCGV Stack (1996-1999) as chief unix engineer, president of the board, verification committee member.
I obtained a Bachelors degree cum laude in Nov 2002 with a 93% score, with an award for Best Achievement in my graduation year. My paper and presentation, written in LaTeX, are available here (Dutch only):
Professional experienceI consider myself an allround engineer, with a solid mix of network-, software- and release engineering and design. But above all, I am a seasoned unix administrator.
I am capable of interpreting and understanding internet standards (RFCs) as well as corporate technical and process documentation (and often use them as a guide in my day to day work) and consciously make make or buy decisions.
Design and Execution
- Network - I took BIT bv from an E3/STM1 Cisco based network to a redundant GigE Juniper based network, and connected it to various european IXPs (DE-CIX, BNIX, AMS-IX, NLIX, GNIX), implementing the peering policy and automating monitoring in the network operating center.
- ISP Subsystems - I have authored process and provisioning systems in the ISP domain (dns/drs, http, smtp/imap/pop3 and radius/ldap/dhcp), using ITIL concepts creating a fully working provisioning database for a 10000+ customer ISP.
- High Availability - I have initimately worked with Google's server HA systems (mostly in-house developed), achieving many-nines uptime and availability for google maps and earth and websearch backends. Using open source principles, I have built PaP Hosting, an off-the-shelf web-, name- and mail-server cluster.
- Very Large Scale Computing - At BIT the IT infrastructure spanned O(100) machines, for which I wrote provisioning and versioning software. At Google, the IT infrastructure I was directly responsible for spanned over O(100K) machines, at which scale compute utilization is key. I drove several utilization and latency projects, saving millions in operational expense.
ManagementThe executive board at BIT BV asked me to take lead of the Engineering Department (12 FTE) in June 2004, during a reorganisation. I performed this task ad interim for one year. At Google, I became an manager for the Geo SRE team in May 2007 (15 FTE), and of the broader Search strategic area in November 2009.
I have developed soft- and technical leadership skills, and visited introspecting and group dynamics courses.
Coaching and Mentoring
- Expert sh(1)
- Good SQL (MySQL, PostgreSQL) and LDAP (OpenLDAP)
- Moderate perl, C / C++, tcl, python
- DAPd (systems monitor used to keep a good eye on the unix servers at large scale deployments such as ISPs)
- YAPS (PCAP based packet accounting tool)
- YANA and 6TRAF (netflow accounting tools)
- ASCC (BGP+RIR dnsbl zonefile generator)
- PXE FreeBSD & Linux (Standalone initrd/kernel for diskless unix servers)
Network EngineeringExtensive design, operational and troubleshooting skill in TCP/IP networks (since 1996). I have been involved with:
- MCGV Stack (TU/Eindhoven university computing club)
- Event networks (Takeover, SANE, HCC Opendagen, Megabit)
- Freeler BV (~1M subscriber free dialup ISP)
- Intouch NV
- BIT BV (WDM network, Extreme L2, Juniper L3)
- A european ccTLD (Cisco L2/L3)
- IXP connection (nlix, amsix, bnix, decix, gnix), peering policy and politics
- Switching protocols, such as EAPS, STP, ATM
- Routing protocols, such as OSPF, ISIS, BGP
- In-depth application protocols: SMTP, POP3, IMAP, HTTP, IRC, FTP, DHCP
- Cisco routers (IOS)
- Juniper routers (JunOS)
- Cisco switches (CatOS/IOS)
- Extreme switches (Extremeware)
- Network Appliance filers (OnTAP)
- F5 loadbalancers (LTM v9)
Unix Systems Administration
- Expert in reliability, stability, performance and scalability issues of Unix systems
- Expert in Linux distributions: Slackware, Debian
- Expert in FreeBSD (4.x, 6.x)
- Moderate in NetBSD/OpenBSD
- Moderate in Mac OSX
- Moderate in Linux Kernel
- Poor Solaris, HP/UX, Windows
Organising an Event/CongressI have acted as lead organizer in two events:
- SANE (program committee)
- NLUUG (program committee)
- What The Hack (program committee)
- Megabit 2002-2004 (network design)
Teaching/SpeakingI have tought TCP/IP (IPv4 and IPv6), Linux and other courses for several companies, and have presented on many technical topics:
- IOC - commercial TCP/IP and Linux training
- FDDI Belgium - commercial IPv6 training
- IPv6, DNS/DNSSEC, RIPE meetings
Current Employment - GoogleI started at Google in August of 2006 as an SRE - Site Reliability Engineer. I chose the Zurich office to work from, and relocated to Switzerland in November of 2006. In this function, I was allocated to the Geo properties, which are rather well known Google Maps and Google Earth. SRE is the group which has the responsibility of user-facing site and all of the infrastructure behind it. Google brings serving to a whole new level, using tens of thousands of (cheap) machines and a unique way of solving large scale computing problems such as storage and serving capacity issues.
In May of 2007, I was asked to assume management responsibility for the Maps SREs. In this new function I take care of the interaction with other groups within the company, such as software engineering groups (spread over 10+ locations), release engineering and product/feature launches from the reliability and serving quality angle. I am proud to note that managers at Google are predominantly engineers maintaining a strong feeling for the technology they work with, as well as the interpersonal communications that make a company of this magnitude scale well. I still carry a pager and carry out 'regular' engineering work.
BIT BV (Febuary 2002 - August 2006)BIT BV is a business oriented ISP based in the Netherlands. With an extensive reseller network of IP related services, the customer base exceeds 15000 companies and 1200 resellers. As part of the engineering department, I am responsible for physical network infrastructure from Ede (Gelderland) to the various internet exchange points (AMSIX, NLIX, BNIX, GNIX and DECIX) connected by a five point ethernet ring spanning 260 kilometers in privately owned dark fiber. I developed a new generation core network based on Extreme Networks layer2 ethernet switches and Juniper m-series IP routers for IP data, and Lucent AM1+ SDH nodes for ATM and voice traffic. To facilitate both telephony and data over the dark fiber, we make use of CWDM transmission equipment to enable up to eight seperate SDH/Ethernet channels. Next to the network deployment, software engineering remains an important part of my duties. I've developed several open and closed source programs, such as YAPS and YANA, BBSMTP (the BIT Batched SMTP Server, a program that sends POP mailboxes to endsites via SMTP), PSA (a PHP/SQL based shell- and webDNS management system) and several other local ISP (sub)systems (such as HTTP and SMTP provisioning systems). I am combining the two disciplines in my current role. I have designed or supervised the following projects:
- PXE FreeBSD4 and FreeBSD6 distribution for clustered applications
- A set of two high availability and performance F5 BigIP v9 loadbalancers (IPv4 and IPv6)
- An IMAPs and webmail servercluster for 3K users
- An SMTP servercluster with AUTH, TLS, SSL, IPv4 and IPv6, handling 100K mails/day
- An MX servercluster with TLS, SSL, IPv4 and IPv6, handling 1.5M mails/day
- Mailscanner servercluster for 500K mails/day
- Apache2 webserver cluster (8 nodes) for 1K sites and 50M hits/day, with a custom suexec, IPv4 and IPv6 support and clever log sorting and distribution
- Design and turn-key implementation of a european ccTLD's core network and application loadbalancing (WDM network, Cisco switches/routers and a cluster of F5 loadbalancers)
- IPv6 deployment of all critical services using IPv6: IP routing, our Unix and Dotnet webfarms, DNS (resolvers as well as authoritative nameservers), FTP (upload and mirrors), SMTP (inbound and outbound), NTP (stratum 1 and 2 servers), NNTP (feeders). BIT pioneers in IPv6.
- A DNSSEC aware plugin for PSA - a CLI and GUI based front- and backend for our authoritative and secondary nameservers; together with a large scale preproduction test scenario involving 15.000 domains within the .nl ccTLD.
WiseGuys (January 2000 - Febuary 2002)Formed December 1999, this software engineering company dedicates its efforts to datamining and information retrieval on the Internet. Offering leading technology searchengines for the mobile industry (WAP) as well as the conventional webengines (HTML), WiseGuys develops and exploits their own datamining software. I was solely responsible for the fault-clearing service, which is a 24/7 pager duty; It required an intimate knowledge of the system software and applications involved. I documented most software and hardware at the company (consisting of approximately 10 servers) and also maintained the network, so I was able to guarantee a high degree of freedom for WiseGuys employees as well as a stable environment for them to work in. I developed two iterations of distributed webcrawlers for the company (including HTTP retrieval, tokenizing and link-extraction and storing of HTML data on-disk). I left WiseGuys when I started my internship for a bachelors degree at BIT BV.
Freeler (August 2000 - August 2001)Freeler, a large scale Free ISP in the Netherlands, has some 600.000 active customers and around 1.300.000 customers in total. This ranks the company in the top percentile of Europe. Running the technical aspects of this ISP, the Informations-Lab (Ilab) consisted of 9 FTE. I was responsible for the physical network infrastructure at the AMS-IX on three locations; Telecity, Nikhef and Atenlabs. Redesigning the core routers, renumbering the networks to sustain the growth of the providers infrastructure and maintaining a decent basis for the other departments to run upon, were my main tasks at this provider. Next to that, I reimplemented the DNS subsystem and ran shifts of fault-clearing duties. Freeler terminated all employee contracts in a merger/sale to Energis.
Track/Netcast/Wegener eMedia (December 1996 - January 2000)Starting in 1996, Toine Verheul was working on an idea to implement a fulltext retrieval and search system for the Dutch Web. I was his first employee back then. We had an agreement with the KUB (University of Brabant, NL) that we'd write the software in exchange for facilities such as hosting. My tasks in the (small) company were maintaining the serverbase (two or three Linux boxes at that time) and coding the webspider system. To date, I have coded four of such spiders, one of which was the central theme for my Fontys College internship for which I got a remarkably high grade (90%). Verheul sold the company to Wegener Arcade, which merged us into Netcast, and later on through a series of companies in the holding, such as Wegener Online Publishing (WOP) and Wegener Electronic Media (WEM). I left Wegener when WiseGuys was founded.
Philips C&P (Febuary 1996 - December 1996)My daily chores consisted of implementing CRs (Change Requests). This basically ment installing software on workstations, installing new laptops and workstations for the employees of Philips Nederland, and maintaining and troubleshooting network and account problems with Novell servers. Under Aad van Jaarsveld, we migrated some 600 employees in the Philips Nederland complex (Vredeoord, Eindhoven) from their WordPerfect/Lotus123 environment to the newer MS Office environment, which was imposed as corporate policy when Philips C&P merged with BSO Origin.
Akuaba (June 1995 - June 1996)Akauba is a private medical company owned by a dr. M.A.H.M. Wiegerink (MD, PhD), a gynacologist, who currently works at the Sint Jozef Hospital [Q3 2003 - Maxima Medisch Centrum] in Veldhoven. Doctor Wiegerink is an infertility specialist, who counsils couples that are having difficulties getting children. I worked together with him developing Progno in 1996, an MS-DOS based program that computes the chances of these couples' spontaneous pregnancy, as well as their chances with several treatments, such as In Vitro Fertilisation (IVF) and Intra Uterine Insemination (IUI), with or without Clomephene treatment. The program was taken over by Organon Medical, and put into care of the World Health Organisation, after which I stopped development on it.
Hobbies/InterestsWhat I enjoy doing in my free time is, amongst others:
- Watch movies in my own private home movie theater (DTS!)
- Play eight and nine ball pool (American style)
- Play with my car stereo (yes, it's big)
- Go on short trips; London, Stockholm, Wroclow, Milan, Rome, Pisa, New York, Dublin
- Combine chatting and trips: Meet people I know from the IRC
- Training two to three times a week at the local cardio gym
- Cooking and enjoying fine cuisine
- Drinking Scottish Single Malt Whisky