150 likes | 288 Vues
This report outlines significant networking upgrades at the Thomas Jefferson National Accelerator Facility, including the transition to a 10Gb Metropolitan Area Network (MAN) with connections to ESNet and NLR. It highlights the implementation of secure wireless networking using WPA, alongside various operating systems like Windows XP, RHEL, and Mac OS X. The report also discusses enhancements to email servers, file storage solutions, and the integration of advanced systems like JASMine and Auger for job management. Issues with power, cooling, and infrastructure monitoring are also addressed.
E N D
Jefferson LabSite Report Kelvin Edwards Thomas Jefferson National Accelerator Facility HEPiX – Fall, 2005
Networking • WAN Upgrade • Upgrading to 10Gb MAN with connectivity to ESNet and NLR • Wireless • Implementing secure wireless using WPA • Working with Windows XP SP2, RHEL3/4, MAC OS-X • WLSE installed for management and to detect rogue access points • Looking at AirDefense for better rogue access point detection and IDS • VLans • Provides functional vs. physical network segmentation
Central Computing • Email • Installed and configured a secure email server • Upgraded our SMTP email hardware for better performance and failover • Examining Solaris 10 zones • Lightweight services placed onto a single machine which appears as two
Central Computing (2) • RedHat EL3 and EL4 • EL4 used for newer servers • EL3 used for desktops and farm nodes • RedHat Network Satellite • Currently at version 3.7 • Upgrading to version 4.0 • Provisioning support • Solaris patch support
Central Computing (3) • Windows builds • New builds get Windows XP SP2 installed • Evaluating the use of Folder Redirection for storing desktop files onto a central server (MyDocuments, etc) • Symantec Client Security • Upgraded from Symantec AntiVirus Corporate Edition • Includes malware detection and removal • Includes firewall, but we’ve disabled • All of this is manageable via a central console
File Server Storage • Installed a 25TB Panasas system • Working to resolve a few minor issues • Memory problem with automount of DF client • Access time was a big issue for us • Finally resolved with version 2.3.1 and pan_atime client • Installed 2 StorageTek B280 systems (30TB) • Fiber Channel disks and controllers • Using these for NFS file service • Very reliable and stable
File Server Storage (2) • Evaluated StorageTek Flexline B680 system • Similar to B280, but uses SATA drives • Not yet ready for production • Looking for an inexpensive, low maintenance Unix-based solution for NFS with reasonable throughput
JASMine Upgrade • Centralized intelligent dispatcher installed • Increases throughput • Small file bundling • Reduces load on the database • File size limit increased from 2GB to 20GB • Supports tape reuse • Copying/compressing data from 60GB 9940A to 200GB 9940B drives • 5000 tapes to be reused at $80/tape
JASMine and Auger interaction • Auger is JLab’s batch farm management system • Tightly integrated with JASMine • Share/reference a common MySQL database • Smart data staging for farm jobs
Grid Developments • PPDG Storage Resource Manager developers meeting at JLab in Sept • revisit SRM requirements document • JLab has SRMv2 service, SRMv3 prototype
Infrastructure • Power/Cooling issues • Problems with current Generator/UPS systems • Hot Aisle/Cold Aisle philosophy for new computer room • Location of Air Conditioning thermostats
Infrastructure (2) • SiteView software • Provides an ability to drill down to see Air Conditioning and UPS status in near real-time. • Provides alarms if values exceed set thresholds • Viewable from web, on and off site.