1 / 18

Experience with procuring, deploying and maintaining hardware at remote co-location centre

Experience with procuring, deploying and maintaining hardware at remote co-location centre. CHEP’13 14 th October 2013

jewel
Télécharger la présentation

Experience with procuring, deploying and maintaining hardware at remote co-location centre

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Experience with procuring, deploying and maintaining hardware at remote co-location centre CHEP’13 14th October 2013 Afroditi XAFI, Alain GENTIT, Anthony GROSSIR, Benoit CLEMENT, Eric BONFILLOU, Liviu VALSAN (since May 2013), Miguel COELHO DOS SANTOS, Olof BÄRRING, Vincent DORE, Wayne SALTER

  2. Outline • Background: why remote co-location? • Preparation • Hands-on access: why, when, who? • First remote deployment • Ramp up remote operation • Conclusions Experience with procuring, deploying and maintaining hardware at remote co-location centre - 2

  3. Background 2012 2011 2010 2009 2013 2007 2006 2008 Contract signed with Wigner Data Centre in Budapest (*) Tender for contract for co-location up to 2.5MW equipment for a duration 3+1+1+1+1 years First deployment: 400 servers 80 disk arrays (6PB) Start with local area co-location: 17 racks, <100kW safe power Construction Exercise ‘remote’ operation In local co-location Containers? New DC? Updated forecast: exceed available power in ~2012-13 Initial forecast predicted exceeding available power (2.5MW) in ~2010 (*) See http://indico.cern.ch/contributionDisplay.py?contribId=1&sessionId=1&confId=220443 Experience with procuring, deploying and maintaining hardware at remote co-location centre - 3

  4. Preparation • Review main processes • Delivery requirements • Hardware handling • Stock management • Inventory • Networkregistration • Burn-in • Production deployment • Remote console • Onsitemaintenance Experience with procuring, deploying and maintaining hardware at remote co-location centre - 4

  5. Preparation • Review main processes • Delivery requirements • Hardware handling • Stock management • Inventory • Network registration • Burn-in • Production deployment • Remote console • Onsite maintenance Experience with procuring, deploying and maintaining hardware at remote co-location centre - 5

  6. Delivery requirements • Delivery requirements stipulated in RFP spec and purchase order: • Firmware versions & settings • Labeling stickers (s/n, MAC, IO ports, disk,, ...) • Wrong settings tend to break procedures and automation • boot order, NIC with PXE disabled, … • struggle with suppliers • Remote console access before delivery • Check and determine detailed settings for supplier • Custom barcode stickers Experience with procuring, deploying and maintaining hardware at remote co-location centre - 6

  7. Network registration • Custom Asset Identifier set by supplier • FRU attributes in BMC • Contract number in ‘Product Asset Tag’ (PAT) • Serial number in ‘Product Serial’ (PS) • On chassis • Bar-code sticker “PAT-PS” • Network registration: • Host generates its name from Asset ID in BMC • Asset ID too long for Windows NETBIOS name. Compromise: • Host name – Asset id association is stored in the network registration database (LANDB) Experience with procuring, deploying and maintaining hardware at remote co-location centre - 7

  8. Automated registration Register ‘p0123456789abcd’ Register asset info DHCP PXEboot Discover MAC addresses HW Discovery Start burn-in Permanent IP Temporary IP HW Inventory Network DB Load Live image Experience with procuring, deploying and maintaining hardware at remote co-location centre - 8

  9. Burn-in & performance tests • Runs as part of the live (in memory) image • Memory (memtest) and CPU (burnK7 or burnP6, and burn MMX) endurance tests • Disks endurance tests (badblocks) • CPU and Disk performance tests (HEP-SPEC06, FIO) • Network endurance & performance tests (netperf) currently require manual start-up … traced to wrong BIOS settings HEP-SPEC06 too low (expected >280) Experience with procuring, deploying and maintaining hardware at remote co-location centre - 9

  10. Automation • Guiding principles: • If some process can be fully described in a manual procedure… • …it might also be scripted • Not always worthwhile in short term • Resilience is paramount • Failures unavoidable and usual require manual action • BUT, it might be possible to carry-on anyway Experience with procuring, deploying and maintaining hardware at remote co-location centre - 10

  11. Onsite maintenance <2012 CERN Supply contracts Repair tickets Repair tickets Repair tickets … Supplier A Supplier B Supplier C Service contracts Service company X Service company Y Service company Z … Experience with procuring, deploying and maintaining hardware at remote co-location centre - 11

  12. Onsite maintenance 2012 CERN Supply contracts Supplier A Supplier B Supplier C … Failed/replace parts shipping Service company Repair tickets Stock mgmt Stock of spares (part of supply) Contract with one service company at each location (Geneva, Budapest) Service contract Experience with procuring, deploying and maintaining hardware at remote co-location centre - 12

  13. Hands-on access • Why, When, Who? • In principle only to rack mount, cable and repair • But… • Remote console missing or not enough • BMC stuck or remote access not working • Use of switched PDUs helps • Wrong settings • improve delivery process! • Cabling • Unavoidable! • Risk mitigation: • Label with i/o ports • Cabling diagrams • Resilient automation • Restricting physical access • Unpopular but unclear why… • … say hello? Experience with procuring, deploying and maintaining hardware at remote co-location centre - 13

  14. First remote deployment • In autumn 2012 we sent out two RFPs • Servers: 300’000 HEP-SPEC06 • Storage: 28PB raw disk in JBODs • Delivery to CERN and Wigner • Two supply contracts per tender • Deliver 35% of servers to Wigner • Deliver 25% of storage to Wigner • + stock of spare parts for on-site repairs Experience with procuring, deploying and maintaining hardware at remote co-location centre - 14

  15. Power on Experience with procuring, deploying and maintaining hardware at remote co-location centre - 15

  16. Registration + burn-in • Power up 400 servers + 80 JBODs • Ran un-assisted • Whole process completed in 2 weeks • 99% success Experience with procuring, deploying and maintaining hardware at remote co-location centre - 16

  17. Status ofremote operation • Hardware handling • Delivery notification • VAT exemption • Goods reception • Scan bar codes • Inventory • Rack mounting • Hardware repair • Notification tickets (Service Now) • Training and documentation • Scheduling • Stock management (Infor EAM) Starting now Experience with procuring, deploying and maintaining hardware at remote co-location centre - 17

  18. Conclusions • Remote co-location is our way to scale beyond local power limitation • Wigner contract awarded following competitive tender • Preparation had positive impact also on local operation • Design workflows and automation with remote operation in mind • Production service is up and running • But work still required to finalise operational procedures • Started preparations for large scale (90%) deployment of new deliveries in 2014-15 Experience with procuring, deploying and maintaining hardware at remote co-location centre - 18

More Related