Hardvér pre vyhľadávanie a indexácia záznamov pre potreby projektu Národnej vedeckej knižnice

null

The core scope of work for this tender involves the delivery of hardware along with comprehensive implementation and service services, totaling 50 man-days (MD). These services include the responsibility for the operation of the delivered IT infrastructure.

The implementation services cover:

• Initial environment analysis, deployment design, and a detailed implementation plan (including rack layout, power, IP plan, RAID, and partition scheme).
• Physical installation of hardware into a 19" rack, including sliding rails and power/data cabling (100 Gbps QSFP, 10/25 Gbps SFP).
• Configuration of firmware/BIOS, RAID controllers, and remote management (BMC, DMTF Redfish API), along with Secure Boot, TPM 2.0 settings, and security hardening according to NIST SP 800-147B.
• Installation and configuration of OS/hypervisor on server types 1 and 2.
• For AI/ML servers: OS installation, configuration of the GPU stack (NVIDIA driver, CUDA, possibly NCCL/GPU fabric), and validation/load testing of all supplied GPUs.
• Network integration into the existing client environment (VLAN, routing, 100/25/10 Gbps connection).
• Integration into client support services: monitoring, backup, AD/LDAP, DNS, NTP, and audit logs.
• Functional and performance acceptance tests, culminating in an acceptance protocol.
• Preparation of documentation (design, as-built, operational).
• Knowledge transfer and training for the client's administrators.

Service services supplement the warranty and include operational tasks not covered by the warranty:

• Preventive maintenance and managed application of firmware and driver updates during planned service windows.
• Periodic health checks of servers, RAID arrays, disks, and GPUs.
• Technical support (L2/L3) and coordination of hardware fault escalation with the manufacturer, including cooperation in warranty claims.

The supplier is also responsible for the proper operational capability of the delivered hardware infrastructure during a six-month stabilization period from the signing of the acceptance protocol. This responsibility includes monitoring the functionality of the delivered hardware, assisting in incident resolution, and restoring the operational capability of the delivered hardware, but it does not cover the operation of the client's applications and data. Service Level Agreements (SLA) for reporting faults are 9x5, with a reaction time no later than the next business day. Unused man-days can be utilized for development and configuration tasks related to the delivered infrastructure during the contract period.