Supply Of Scientific Equipment: 1 Master Node: 1 No A 2 { Intel Xeon E5-2630V3 8C 2.4Ghz 20M 8Gt / S B 64 Gb ( 16Gb X 4 ) Ddr4, 2133 Mhz Ecc Memory, In Balanced Configuration For Maximum Performance C 6 × 4Tb Sata Hard Disk @ 7200 Rpm Or Better With Hardware Raid With Support For 0, 1 And 5 / 6. D Four Gigabit Ethernet Ports With Pxe Boot Capability E Dvd-Rw Drive F Ipmi 2.0 Or Equivalent Support With Kvm And Media Over Lan Features. Must Include Any Licenses, If Required For Using All Features. G Fully Compatible / Certified With Rhel6.X, Rhel7.X And Suse Linux Operating System. H 2U Rack Mountable Or Smaller Form Factor. I Redundant And Hot Pluggable, 80 Plus Platinum Or Better Certified Power Supply. 2 Compute Nodes: 4 Nos A 2 { Intel Xeon E5-2630V3 8C 2.4Ghz 20M 8Gt / S B 64 Gb ( 16Gb X 4 ) Ddr4, 2133 Mhz Ecc Memory, In Balanced Configuration For Maximum Performance C 500Gb 7.2K Rpm Sata Disks D Two Gigabit Ethernet Ports With Pxe Boot Capability. E Ipmi 2.0 Or Equivalent Support With Kvm And Media Over Lan Features. Must Include Any Licenses, If Required For Using All Features. F Fully Compatible / Certified With Rhel6.X, Rhel7.X And Suse Linux Operating System. G 1U Rack Mountable Or Smaller Form Factor. H Redundant And Hot Pluggable, 80 Plus Platinum Or Better Certified Power Supply. 3 Network / Interconnect A Managed Gigabit Ethernet Switches ( 24 Ports Minimum ) To Connect All Hpc Nodes For Cluster Management, Provisioning And Ipmi Management. B Required Numbers Of Cat6 Molded Appropriate Length To Suite The Solution 4 Rack And Accessories A One Rack With Proper Arms And Channels, Power Points And Proper Perforated Doors ( Front And Back Both Side ) With Cooling Arrangement Etc., For Servers Sufficient To Accommodate At Least 5 Servers And Network Equipments, Kvm Console And Kvm Switch And Other Associated Accessories Etc. B Kvm Switch ( Quantity -1No ) Compatible To The Servers Offered With 8 Ports Console Cables Required: 4Nos. Of 2 Meter Dual Ps2 / Usb Along With Vga Cables Required For Connecting 8 Systems Cables, Console Interface: Accepts Either Ps / 2 Or Usb Keyboards And Mice. C Console: Rack Foldable 17 Tft In 2U / 3U Form Factor With Mouse And Keyboard 5 Software A System Software I Operating System Can Be Open Source 64 Bitlinux B Compilers And Libraries I Openmp, Mpi, Gnu C / C++ And Fortran Compilers Should Be Installed. Ii Open Source Libraries Lapack, Blas, Fftw Etc C Applications That Will Be Run: I Please Add The Applications D Cluster Management Software ( Can Be Open Source ) I Web-Based Application For Supercomputer Access. Platform Independent Interface ( Windows / Unix / Mac ) . Gui Monitor Should Support All Browsers. Ii User Account Management From Master Node. Iii Monitoring Tool Like Ganglia Or Better, Integrated With Environment Values, Load, Resource Status, System Environment Status, Hard Disk Usage Status And Hardware Configuration. Iv Users’ Home Directory Should Be In File System With Quota Enabled. V Reimaging New Nodes ( Cpu ) From Master Nodes. Vi Integration Of All Software Components So As To Make The Complete Hpc Cluster System Fully Functional And Usable ( E.G., Integration Of The Scheduler With Mpi, Any License Managers, Etc. ) . Vii Detailed Reports About Cluster Usage Statistics. Reports Of Every User And Jobs Including Their Monthly Usage, Node Usage, Percentage Of Utilization, History Tracking, Number Of Completed, Failed, Queued And Runningjobs; Estimated Delay And Average Job Duration. Viii Automated Health Check Of Nodes And Urgent Alert Messages About Critical Errors. E Job / Work Load Management ( Can Be Open Source Must Comply Atleast Following Points ) I Jobs Can Be Submitted From Master Node Only. Ii Cpu Enabled Scheduling With Checkpoint And Restart. Iii Job Scheduler Configuration Including New / Delete / Disable Queues. Iv Dynamic Resource Management And Resource Balancing. V Policy Based Resource Allocation. Vi Configure Job Scheduler For Script Less ( Short Command Line ) Submission Master Node. Vii Job Status Monitoring, Job History Viii Usage Accounting & Reporting With Total Utilization Of Resource. Ix Check Runaway Process Before Entering New Jobs. F Firmware I All Hardware Should Be Installed With Recent Stable Version Of Firmware. 6 Warranty And Support ( Not Optional ) A Cluster Management And Support For 3 Years – B Training For General System Administration With Documentation Including As User / Node Management, Installation / Upgrade, Queuing System Management File System Management. C One L2 / L3 Level Trained Personnel Should Be Available To Help At Any Time Remotely Or In Person. D Technical Support For Administration / Maintenance ( Both Software And Hardware Levels ) Of Hpc. Vendor Will Be Responsible To Protect Data During Any Upgrades Firmware / Os. E A Helpdesk Email Account Which Is Regularly Monitored Should Be Available Users. F An Escalation Matrix For Issues Not Resolved By The Support Personnel, With Expected Time Line, Should Be Clearly Mentioned. G The Person Should Have Enough Experience To Handle Cluster Hardware Troubleshooting To Resolve The Problems Faced By The Users. This Should Tuning Of The Scheduler’S Various Capabilities. H The Person Should Be Able To Produce Required Status Report Of The Cluster Asked Using The Software Installed In The Cluster To Manage It. I Faulty Parts Should Be Replaced Within 48 Hours Of Logging A Call. J Hardware Warranty For 3 Years. K 3-Years Of On-Site Comprehensive Warranty. Mention Annual Maintenance ( Amc ) Charges For 4Th & 5Th Years Separately ( Optional, Quote Compulsory ) . 7 Documentation A User Creation / Deletion / Modification. B Bringing Up And Shutting Down The Cluster. C Basic Troubleshooting For Raid Storage ( Master Node ) And Job Scheduler. D Step By Step Installation Guide For Node Configuration From Scratch. E When Handing Over The Cluster The Vendor Should Provide The Full Design Of The Cluster Installation Including The Electric Connections, Network Connections, User Manual Clearly Explaining How To Use The Cluster. 8 Scope Of Work With Deliverables To Be Part Of Implementation: A Physical Verifying Hardware Items In Bill Of Materials B Rack Mounting All The Hardwares And Connecting Power Cables & Ethernet Cables To All The Nodes C Hardware Installation Of Master, Compute Nodes, Switches, Console In Rack, Implementation & Configuration D Installation And Configuration Of Cluster Os And Cluster Toolkit On Master Node And Compute Nodes E Installation And Configuration Of Scheduler, Queues, Users And Policies ; Policies To Be Discussed With Jmi Before Installation Process And Implemented Installation Of Compilers & Libraries F Installation And Integration Of Open Source Compilers And Libraries G Integration Mpi And Schedulers With Compilers H Mapping Internal Strorage Luns To Cluster I Configuration Of Luns On Cluster According To Requirement J Testing And Verification Of Complete Setup Functionality Applications K Applications : Installation And Configuration Of Applications L Benchmarking: Demonstration Of High Performance Linpack ( Hpl ) Benchmark Performance Of Minimum 75Perc. . M Training : Cluster Usage Training And Scheduler Training To End Users – Good And Nicely Written User Manual On How To Use The Cluster ( Documents Should Readable By A New User Without Much Help ) . Demonstration On Details Of The Scheduler And Other Necessary Cluster Related Software Packages. Remote Or In-Person Help For New User To Get Used To The Cluster And Troubleshooting Of Problems Faced By Different Users Immediately. N Documentation And Training– A Detailed Document About The Cluster Including Hardware And Software Details. O Project Final Signoff
