Report for April 2022

Report Date

April 2022

Achievements in Last Period

• Eddie/Eleanor nodes (14 x C6400 chassis x 4 x C6420) have been delivered to OCF and are in stress test/burn-in (PO IS22125/IS22122)

  • Burn-in completion estimated 09/05/2022
• DataStore upgrades (2 x 2 x ME4084, 15 x R640) are on site at JCMB (PO IS2121/IS22117)
  • Installation of storage completed at JCMB/AT on 14-15/04/2022
• TSM storage arrays were installed at ACF on 13/04/2022 (PO IS22115)
  • Date for TSM array configuration needed
• Eddie Dell storage servers (4 x R640) to be shipped direct to ACF on 05/05/2022 for installation on 06/05/2022 (PO IS22123)
  • RSS team onsite on 05/05/2022 to receive delivery and install on 06/06/2022
  • RFCs raised for servers and storage
• Eddie 7 x Lenovo DE6000 storage; shipment (one pallet) delivered to ACF 25/04/2022 (PO IS22123)
  • Partial delivery, follow up with supplier
• Orders placed on 22/04/2022 and 26/04/2022 for current quotes
  • PO IS22500 raised for quote 1003673743 for 2 x ME4024
  • PO IS22487 raised for quote 1003584021 for 3 x R650 server
  • PO IS22501 raised for quote 1003671671 for 18 X N1524 switches
  • PO IS22493 raised for quote 1003673974 for 4 x 10/25GbE Mellanox network cards
  • PO IS22488 raised for quote 1003673974 for 2 x 100GbE Mellanox network cards
  • PO IS22491 raised for quote 1003675841 for 1 x R640 server (Globus/Eddie)

Risks/Issues

• Lack of information on order status from Esteem has meant that additional RSS time has been spent clarifying status (-)

  • Status of all original orders now known and being tracked (+)
  • Orders placed for all remaining quotes (+)
  • Weekly review meetings now in progress with Esteem/Dell/TechData (+)
• Replacement needed for servers and storage approaching EoL (plus need for additional processing capacity) (-/+)
  • DataStore kit currently EoL with more in July/August (-)
• RSS resource capacity is at risk as project (P1) will run in parallel with RSS401 GPU Refresh (P2) and RSS405 Tape Library (P1) (-)
  • Aggregate planning has identified dates that avoid other critical work (+)
• Current allocated budget is £2.258M and spend must be within tolerance (+/-)
  • £7K projected overspend approved (+)
  • Dell price increases at end-May may impact planned spend (-)
• Equipment lead times may delay project completion past July 2022 (-)
  • Bulk of orders will be received by July 2022 (+)
• Installation and/or configuration must be carefully coordinated with ACF (RFCs) and ISG (GoCAB) (+/-)
  • ISG GoCAB requests and ACF RFCs being raised as required (+)
• Upgrade work will impact DataStore, Eddie and Eleanor services (-)
  • DataStore server configuration will mean momentary outages (-)
  • Reduced Eddie/Eleanor service when compute nodes are installed and configured (-)
• Planned kit may exceed data centres’ power budget (-)
  • Reviewing server/storage ratings against power thresholds (-)
  • No issue at JCMB/AT; estimated as a reduction in ACF (+)
• No space in racks at ACF for additional kit (-)
  • Swaps out/in ACF will mean reduced service (-)
  • Power densities and racking impact considered low risk (AT)
• OCF, StackHPC and IBM resources may not be available for installation and configuration dates (-)
  • Dates for Eddie/Eleanor nodes installation/configuration moved to week 01/08/2022 at request of OCF (-)
• Storage areas at ACF and JCMB need to be agreed and storage time minimised (-)
  • ACF storage currently holding decommissioned tape library kit; due for to be removed 04/05/2022 (-)
• COVID restrictions at ACF and JCMB/AT may restrict access and so impact dates (-)
  • Restrictions have been remove at ACF/JCMB/AT
• UCU industrial action may impact project progress (-)
  • Potential further action at end-May (-)

Next Steps

• Complete installation of DataStore servers at JCMB

  • Confirm racking date for DataStore R640s at JCMB with Esteem/Dell
  • Planned configuration date for servers 24/06/2022
• Configuration date needed for TSM storage arrays at ACF; not in current plan
• Delivery of Eddie Dell storage servers to ACF scheduled for 05/05/2022
  • Dependent on removal of old tape library kit by CCL North on 04/05/2022
  • Installation of Eddie Dell storage servers planned for 06/05/2022 by RSS team
• Query status of rest of Eddie Lenovo storage delivery; review planned installation date of 06/05/2022; at risk
  • Arrange IBM support for Lenovo storage installation
• Place any final orders by 13/05/2022
  • Review current order costs versus budget
• Continue installation planning at ACF including onsite storage requirements
  • Confirm status of RFCs for Eddie Dell storage servers/storage
  • Raise RFC for Eddie/Eleanor compute nodes
  • Check with ACF at monthly meeting on earlier delivery date (July) for compute nodes
• Review needed for extended support for DataStore kit currently EOL
• Complete overall RSS resourcing planning in line with the aggregate plan
  • Confirm RSS402 timing and resources

Milestones