Intent to Sole Source
Added: Mar 12, 2018 12:58 pm Solicitation #:N00173-18-Q-JL02Procurement Type:Special NoticeTitle:High Performance Computing ClusterDescription:The U.S. Department of Navy, Naval Research Laboratory (NRL) Contracting Division, proposes to enter into a sole source contract with Hewlett Packard Enterprises Government, LLC (HPEG) for a High Performance Computing Cluster. The minimum required specifications for the High Performance Computing (HPC) Cluster are identified below:One Rack HPC Cluster Requirements1) The HPC cluster must fit within a single rack enclosure.2) The HPC cluster must be composed of compute trays (similar to "blade" technology). Each compute tray must have two (2) processors in a shared memory configuration. The cluster enclosure must contain thirty-two (32) compute trays and support up to forty-eight (48) compute trays without additional infrastructure.3) Each compute tray must contain 2 Intel® Xeon-G® 6136 (3.0GHz/12-core/150W) Gen 10 CPUs and 384 GB Dynamic Random Access Memory (DRAM). The HPC cluster must have a total of 768 cores with 16 GB of DRAM per core.4) All hard disk storage for the compute trays must be Solid-State Drives (SSD). Each compute tray must have one 400GB NVMe SSD.5) The HPC cluster must be factory assembled and tested. It must have full computer system integration and testing before delivery and installation at NRL. It must be shipped to NRL completely ready for immediate deployment only requiring power and network connections. The installation must include on-site technical support and system optimization, technology transfer and training. The cluster must include tee (3) years of maintenance with "next business day" response, CDMR, and all required licensing.6) The HPC cluster must include one (1) head node management server with rack-mount LCD monitor/keyboard/mouse device included. Head node must fit into a 2U slot in the rack and must include the following:a) Two (2) Intel® Xeon-G® 6136 (3.0GHz/12-core/150W) CPUsb) 192GB DRAM interleaved on the motherboard for maximum memory toughputc) Two (2) 600GB SAS 15K HDDs in a Raid1 (mirror RAID redundant) configurationd) Six (6) 8TB SATA 7.2K HDDs Raid6 (can sustain two simultaneous failed drives) configuratione) Smart Array 10G RAID controller with:i) Must have 16 SAS Lanes and 4x4 internal Mini-SAS portsii) Must handle both 12 Gb/s SAS and 6 Gb/s SATA technology HDDsiii) Must support Mix and Match SAS and SATA drives on the same controlleriv) Must support SAS tape drivesv) Must support RAID Levels: 0, 1, 5, 6, 10, 50, 60, 1 ADM and 10 ADM (Advanced Data Mirroring)vi) Must be VMware VSAN certifiedvii) Must support both legacy and UEFI boot operationsviii) Must support up to 238 physical drives/64 logical drivesix) Must support Secure Encryption and full FIPS 140-2 Level 1 compliancex) Must have 96W smart storage batteryxi) Must have 10Gb Ethernet NIC supporting two 562FLR-SFP+Adptxii) Must have HPE ILO4 Management modulexiii) Must have embedded 4x1gb NIC ports xiv) Must have dual, redundant 800W Platinum Power Supplies7) The HPC cluster must include one (1) 48 port 10GbE switch with 952 Mpps/1280 Gbps toughput8) The HPC cluster must include two (2) 24 port 1Gb/s Ethernet switches with 9.5 Mpps/12.8 Gbps toughput9) To fully integrate into our already existing HPE computing environment, the cluster must include the following:a) HPE Integrated Lights Out (ILO) hardware modules toughout and licenses for the ILO management toolsb) Full licensing for the HPE Cluster Management Utility (CMU) version 8 or newerc) Full licensing for the HPE Apollo Platform Manager (APM)d) Full licensing for HPE OneView10) To comply with security best practices and requirements, the cluster must include the following:a) Silicon Root-of-Trust that is burned into the silicon components on the motherboard.b) Must conform to National Institute of Standards and Technology (NIST) security framework known as: "Protect, Detect, and Recovery."11) Cluster must have these rack management featuresa) The cluster must have automatic chassis and device discovery with topographic viewsb) The cluster must t have time-stamped, rack-level event loggingc) The cluster must have rack and chassis shared power and thermal component managementd) The cluster must have integrated serial connector for server console and ancillary device accesse) The cluster must have integrated Gb Ethernet switch for server iLO consolidation12) Cluster must have these server management featuresa) The cluster must have server health monitoringb) The cluster must have iLO Single Sign-On accessc) The cluster must have server FRU data reportingd) The cluster must have iLO IP address and PXE MAC address listings13) Cluster must have these power management featuresa) The cluster must have Power control and measurement at the server, chassis, and rack levelb) The cluster must have PDU-level power outlet control and current measurementc) The cluster must have Rack-level static or dynamic power cappingd) The cluster must have DC Power Shelf management and integration with UPS subsystem14) The vendor must provide cleared personnel for onsite installation, integration, and system tuning.15) The vendor must provide cleared personnel for all onsite system maintenance.High Speed Switch RequirementsTo support cluster operations and user interaction, the following network hardware is required:1) Eight (8) fully port-level managed, 48-port 1Gbs Layer 2 network switches are requireda) Must fully compatible and manageable tough existing HPE OneView environmentb) Must support stacking, fixed 10GbE ports, static layer 3 routing and RIP v1/v2, PoE+, ACLs, IPv6 and power savings with Energy Efficient Ethernet.c) Must provide traffic prioritization with supported congestion actions including: strict priority (SP) queuing, weighted round robin (WRR), weighted random early detection (WRED), and SP+WRR and traffic policing with Committed Access Rate (CAR) and line rateThe contemplated contract will be a Firm Fixed Price (FFP) type action. This notice is not a request for proposals or competitive offers.A determination by the Government to compete, or not, will be based on the responses received, and is solely within the discretion of the Government. Information received will normally be considered only for the purpose of determining whether to conduct a competitive procurement. The FSC is 7025 and the NAICS is 334112 N00173-18-Q-JL02 Department of the Army Materiel Command Army Contracting Command Aberdeen Proving Ground
3/12/18, 12:58 PM High Performance Computing Cluster
-- N00173-18-Q-JL02 Department of the Navy Secretary of the Navy Office of Naval Research
5/15/18, 11:31 AM