Genentech’s IT Infrastructure: Evolving to support the Revolution John “Scooter” Morris, Ph.D. Distinguished Systems Architect Computing Technologies Scooter Morris, CIT (scooter@gene.com) Genentech IT Infrastructure June 12, 2002 page 1 Genentech, Inc. “Genentech is a leading biotechnology company that discovers, develops, manufactures and commercializes biotherapeutics for significant unmet medical needs.” Statistics • ~5,000 Employees • ~$2.2B in Revenue (2001) • 10 products - Protropin®, Nutropin®, NutropinAQ®, NutropinDepot®, Cathflo™ Activase®, Activase®, TNKase™, Pulmozyme®, Herceptin®, Rituxan® • 1 product awaiting FDA approval - Xolair™ • Three major sites - Scooter Morris, CIT (scooter@gene.com) South San Francisco, California Vacaville, California Porriño, Spain Several U.S. Sales offices Genentech IT Infrastructure June 12, 2002 page 2 Genentech, Inc. Founded in 1976 • Herb Boyer – UCSF Professor • Bob Swanson – Entrepreneur, Venture Capitalist Focus areas: • Oncology - 7 drugs or new indications in pipeline - 4 in phase III • Immunology - 5 drugs or new indications in pipeline - 2 in phase III, 1 awaiting approval • Opportunistic - 3 drugs or new indications in pipeline - 1 in phase III Scooter Morris, CIT (scooter@gene.com) Genentech IT Infrastructure June 12, 2002 page 3 Clinical Development of Drugs Discovery Marketing and Line Expansion Development Post marketing studies Synthesis and testing New clinical indications pursued IND plan established and initiated Chemical lead found New dosage forms and formulations developed IND filed Phases I, II, III Additional compounds Clinical studies initiated are made NDA prepared Candidate compound and submitted chosen and additional tests run NDA approved Phase IV Idea for new chemical Compound elevated to project status Safety surveillance Drug launched Scooter Morris, CIT (scooter@gene.com) Genentech IT Infrastructure June 12, 2002 page 4 Outline Current Infrastructure • A bunch of details which I will skip Research Computing Revolution vs. Evolution Evolution of Computing for Research • Evolutionary tree Future Directions Scooter Morris, CIT (scooter@gene.com) Genentech IT Infrastructure June 12, 2002 page 5 Current IT Infrastructure Highly heterogeneous • Servers: SGI, HP Alpha, HP PA-RISC, HP Intel, Sun • Desktops: Mac, PC, some SGI Primarily IP-based network • AppleTalk also supported • Routers and switches: Cisco Security based on M&M principle • Hard outer shell, soft inside • Some “softness” appearing to support collaboration • Important to maintain open environment Scooter Morris, CIT (scooter@gene.com) Genentech IT Infrastructure June 12, 2002 page 6 Current IT Infrastructure Building Building Switch 200 Switch bps 200 M 200 Mbp s Cisco Router 2G bps 200 Mb ps Cisco Router 2 Gbps Cisco Router Cisco Router Cisco Router Cisco Router 2 2G bp s Vacaville 200 Mbps G Switch bp s Switch Switch Cisco Router Switch Switch 2 Gbps Cisco Router s Mbp Cisco Router 2 Gbps Switch Building 4 Computer Room Scooter Morris, CIT (scooter@gene.com) Building 5 Computer Room Genentech IT Infrastructure June 12, 2002 page 7 Current IT Infrastructure Key Firewalls & proxies Web & file sharing VPN DNA DNA Cisco Router Switch Cisco Router Cisco Router genie gnome efreet djinn VPN concentrators Switch Switch Cisco Router Cisco Router Cisco Router UUNET Genuity wallace-ltd outcast2 www-secure outcast Cisco 3002 Partners Partners (frame) (frame) Limited Net Scooter Morris, CIT (scooter@gene.com) Sales Offices Genentech Spain Cisco 3002 Internet Internet Partners Partners (IPSEC) (IPSEC) Open Net Genentech IT Infrastructure June 12, 2002 page 8 Details This starts the part I’m going to leave out…. Scooter Morris, CIT (scooter@gene.com) Genentech IT Infrastructure June 12, 2002 page 9 Details - Network Backbone: Switched Gigabit Ethernet Vacaville link: 200 Mbps SONET Ring Desktop: Switched 10/100 Ethernet Routers: CISCO Addressing: DHCP preferred Naming: DNS (Bind 9.2.1), WINS, Active Directory, LDAP Firewall: SOCKS5 (Aventail) Monitoring: Big Brother, HP Open View VPN: Cisco 3000, IPSEC Scooter Morris, CIT (scooter@gene.com) Genentech IT Infrastructure June 12, 2002 page 10 Details - Desktop Compaq (now HP) • Migrating to Windows 2000 Apple • Migrating to Mac OS X Primary Applications • Microsoft Office 2000/2001 • Netscape Communicator (Browser, Mail) - Migrating to Mozilla/Netscape 7 • Steltor CorporateTime • Norton Antivirus • FileMaker Pro Scooter Morris, CIT (scooter@gene.com) Genentech IT Infrastructure June 12, 2002 page 11 Details - Server HP Tru64 Unix • Web, E-Mail (IMAP), Bioinformatics, Infrastructure (DNS, Firewall, DHCP, backup/restore, LDAP), General computing, Oracle - 5.1A (TruCluster Server 5.1A) HP/UX • Manufacturing, Commercial Computing (Lawson, PeopleSoft) • 10.20, 11.0 Solaris • Medical Affairs, Infrastructure (Calendar, Remedy, Web Proxy), Research SGI • Molecular Modeling, Computational Chemistry Linux • Computational Chemistry (pilot) NT • Workgroup Computing, Specific Applications Scooter Morris, CIT (scooter@gene.com) Genentech IT Infrastructure June 12, 2002 page 12 Details - Database Oracle 8.1.7 • Exploring Oracle 9iRAC n-tier approach • • • • Scooter Morris, CIT (scooter@gene.com) Web Browser for presentation Web server for static pages Application servers for business logic Database server for data store Genentech IT Infrastructure June 12, 2002 page 13 Details - Web Server: Netscape Enterprise Server 4.6 • Migrating to Apache Programming: Perl/CGI, JSP, Javascript Application Servers: • WebLogic Distributed Computing: • Direction is towards Enterprise Java Beans - WebLogic - Tuxedo in use for Manufacturing applications Development Tools: • Dreamweaver, JBuilder, Visual Age Scooter Morris, CIT (scooter@gene.com) Genentech IT Infrastructure June 12, 2002 page 14 Details - Security Security is based on Kerberos V5 Provides authentication for Unix and Windows 2000 Oracle accounts often use Unix username, but also lots of application-specific accounts • Exploring use of Kerberos for Oracle accounts VPN and dial-in access through SecurID tokens LDAP is used for Directory services • Netscape Directory Server Serious regulatory restrictions (21CFR Part 11) Scooter Morris, CIT (scooter@gene.com) Genentech IT Infrastructure June 12, 2002 page 15 Details - Internet/Extranet Firewall is based on SOCKS5 (RFC1928) Totally Proxy-based (very secure) Firewall has three parts: • Internal • Internet • Limited net Internet link is redundant • 42 Mbps link with Genuity • T1 (1.54 Mbps) link with UUNET - Migrating to a redundant 42 Mbps link • Uses OSPF for dynamic fail-over • Using IPSEC tunnels over the Internet for secure communications with partners Cisco 3000 VPN concentrators • Employee access only Reverse-web Proxy • Allows external partners access to selected internal web sites Scooter Morris, CIT (scooter@gene.com) Genentech IT Infrastructure June 12, 2002 page 16 Details Any questions on the details? • I didn’t think so…. Scooter Morris, CIT (scooter@gene.com) Genentech IT Infrastructure June 12, 2002 page 17 The Revolution (a parochial view) Biochemical Cycle Systemics Protein Boyer & Cohen 1973 Itakura & Riggs 1977 Proteomics Human Genome Project 1990 PCR 1983 ESTs 1990 Gene 1978 Insulin Cloned ABI Sequencers 1994 Human Genome Draft 2001 Genomics 1985 Human Growth Hormone Approved 1998 First “humanized” MAB-based Pharmaceutical 1976 Genentech Founded Scooter Morris, CIT (scooter@gene.com) Genentech IT Infrastructure June 12, 2002 page 18 The Revolution More than changes in scale Fundamental changes in scope and complexity • GeneÆGenomeÆProteomeÆSystems Dependent on computation • Dramatic increases: - Performance - Storage capabilities - Communication capabilities Scooter Morris, CIT (scooter@gene.com) Genentech IT Infrastructure June 12, 2002 page 19 Evolution of computing at DNA In the beginning was the…… Apple II Scooter Morris, CIT (scooter@gene.com) Genentech IT Infrastructure June 12, 2002 page 20 Computing “Revolution” (a parochial view) DEC PDP/11 1970 ARPANet 1969 Alpha 1992 Internet 1983 Ethernet 1973 Altair 1975 Apple II 1977 Beowulf 1994 RISC 1986 DEC VAX 11/780 1977 IBM PC 1981 Apple IIe 1983 genie VAX 11/780 1980 Genentech’s First Computer Apple II 1978 genie VAX 8650 1987 56 KB Leased Line to UCSF (TCP/IP) 1986 ruby Alpha 8400 1995 tinman DECSystem 5900 1991 geneland TruCluster 2000 BARRNet Affiliate 1990 1 MIPS 90,000 MIPS MIPS = Meaningless Indicator of Processor Speed Scooter Morris, CIT (scooter@gene.com) Genentech IT Infrastructure June 12, 2002 page 21 Computing “Revolution” Changes in scale • System performance - 1 MIPS Æ 90,000 MIPS (clustered) • Memory capacity/system - 4MB Æ 88GB (~4 orders of magnitude) • Global connectivity - 9.6 Kbps Æ 42 Mbps Changes in complexity • Multiprocessing (threaded code) • Clustering (distributed code) Scooter Morris, CIT (scooter@gene.com) Genentech IT Infrastructure June 12, 2002 page 22 Strategies for IT Change Change is required to: • support changing research needs • take advantage of changing technology Strategies • Revolutionary - Rip and replace - Takes maximum advantage of changing technology • Evolutionary - Incremental changes - Minimizes impact to users - Might result in slower adoption of technology • Genentech has taken an evolutionary approach Scooter Morris, CIT (scooter@gene.com) Genentech IT Infrastructure June 12, 2002 page 23 Setting - Research Goal • Basic research • Human pharmaceuticals Organization (roughly) • Discovery - e.g. Molecular Oncology, Immunology • Technology - e.g. Bioinformatics, Protein Engineering, Bioorganic Chemistry Academic culture • Open, fast-paced environment • Need to provide tools as much as solutions Computational needs are high • Bioinformatics • Molecular modeling • Computational Chemistry Scooter Morris, CIT (scooter@gene.com) Genentech IT Infrastructure June 12, 2002 page 24 Bioinformatics Computing Evolution Disk Subsystem HSG80 HSG80 HSG80 HSG80 FC Switch ES40 GS160 8400 12-processors 10-processors ES40 ES40 ES40 ES40 ES40 ES40 ES40 ES40 ES40 ES40 ES40 ES40 ES40 ruby bioinfoweb MC II Hub Scooter Morris, CIT (scooter@gene.com) ES40 geneland Genentech IT Infrastructure June 12, 2002 page 25 Bioinformatics Cluster Disk Subsystem HSG80 HSG80 HSG80 HSG80 FC Switch trp cytosine guanine thymine cys leu ala met ruby bioinfoweb MC II Hub geneland adenine Scooter Morris, CIT (scooter@gene.com) Genentech IT Infrastructure June 12, 2002 page 26 Protein Engineering / Bioorganic Linux Cluster SGI Origin 2000 (16 processors) SGI Origin 3000 (24 processors) SGI Origin 2000 (12 processors) Scooter Morris, CIT (scooter@gene.com) Genentech IT Infrastructure June 12, 2002 page 27 Evolution of computing at DNA BARRNet UCSF 1980: VAX 11/780 8600 digitaldigital VAXVAX 11/785 digital VAX 8600 • BSD Unix • UUCP Dialup Connection to UCSF 1984: VAX 11/785 • BSD Unix • “Corporate” network 1986: Connection upgrade • 56Kb Leased line (TCP/IP) 1987: VAX 8600 • BSD Unix • gene.com registered • BARRNet affiliate (1990) digital VAX 8600 SGI 4D/220S 1989: VAX 8650 • BSD Unix • Total of 3 shared systems 1989: SGI Server • IRIX • NFS Services • Molecular modeling Scooter Morris, CIT (scooter@gene.com) Genentech IT Infrastructure June 12, 2002 page 28 Evolution of computing GS ph Al X VA 86 DE 50 s te Sy C DE m 5 ys CS t em 70 aS e er rv 84 16 E 0/ -6 4 IA V7 00 00 0 90 M IB SGI 4D/220 ux Lin Fa rm SGI Origin 3000 X-Ray Crystallography X VA 11 / 78 0 1980 1990 VAXocene Scooter Morris, CIT (scooter@gene.com) MIPSocene 2000 Alphacene IPFocene? Genentech IT Infrastructure June 12, 2002 page 29 Evolution of computing – Punctuated Equilibrium? We have been in a period of relative stability Episodes of change are “normal” • • • • Hardware changes OS changes Technological changes (Java, J2EE, etc.) Vendor changes Our job: • Making the change less intrusive - Attractiveness of Java - Vendor stability - Planning • Avoid revolution Scooter Morris, CIT (scooter@gene.com) Genentech IT Infrastructure June 12, 2002 page 30 Futures Scooter Morris, CIT (scooter@gene.com) Genentech IT Infrastructure June 12, 2002 page 31 Linux Pilot Purchased a 24-node Linux cluster • 2 800MHz PIIIs in each node • Myrinet • Computational Chemistry Applications - Amber, Gaussian Results • Memory bandwidth was a problem • PIIIs already out-of-date • Myrinet capabilities not heavily used Hypothesis • Could build a cluster with different types of nodes • Submit jobs to appropriate node depending on computational needs • May not need expensive cluster interconnect Scooter Morris, CIT (scooter@gene.com) Genentech IT Infrastructure June 12, 2002 page 32 Heterogeneous Compute Farm Disk Subsystem HSG80 HSG80 MC II Hub HSG80 HSG80 FC Switch MC II Hub ES45 Gigabit Ethernet P4 Linux IPF Linux/HP-UX Scooter Morris, CIT (scooter@gene.com) Apple Xserve MacOS X Alpha Tru64 Genentech IT Infrastructure June 12, 2002 page 33 Research Computing Environment – Future? Disk Subsystem HSG80 HSG80 HSG80 HSG80 FC Switch trp cytosine thymine guanine leu cys ala met MC II Hub Gigabit Ethernet SGI Scooter Morris, CIT (scooter@gene.com) Genentech IT Infrastructure June 12, 2002 page 34 Questions? Good, I’ve got some questions: • How many are Tru64 customers? • How many are concerned about Itanium transition? - What are your concerns? • How many are concerned about HP/UX transition? - What are your concerns? Scooter Morris, CIT (scooter@gene.com) Genentech IT Infrastructure June 12, 2002 page 35 Thank you! Acknowledgements: Colin Watanabe Carol Morita Roy Takai Pam Fitch Scooter Morris, CIT (scooter@gene.com) Genentech IT Infrastructure June 12, 2002 page 36
© Copyright 2024 Paperzz
