HPE0-S59 Plan and Design Server Solutions

Plan and Design Server Solutions Detailed Explanation

1. Requirement Gathering and Analysis

This is the first and most important step in designing any server solution. Before you choose any hardware or software, you need to fully understand what the customer needs, both technically and from a business perspective.

1.1 Understand Customer Business and Technical Needs

You must ask the right questions and listen carefully to your customer. You’re not just asking what technology they want, but why they need it.

Business Drivers

These are non-technical goals that affect the server design:

• Cost Optimization

  • Can the customer afford the solution?

  • Can we reduce total cost over time?

  • Think about CAPEX (upfront cost) and OPEX (ongoing costs).

• Compliance Requirements

  • For example:

    • GDPR in Europe (data privacy)

    • HIPAA in the US (healthcare data)

  • You may need to store data within certain countries or use encryption.

• Availability / Uptime

  • How much downtime is acceptable? (Usually, the answer is: very little!)

  • This affects choices like redundant power, failover, and high-availability clusters.

• Security and Data Protection

  • Is data encrypted at rest and in transit?

  • Do they need backup and recovery?

  • Can we detect unauthorized access?

Technical Requirements

These are the actual performance and capacity needs of the system.

• Compute Needs

  • Number of CPU cores, threads, and clock speed.

  • How much RAM is required?

  • Use for web servers, databases, etc.

• Storage Requirements

  • Capacity: How much space do they need?

  • IOPS (Input/Output Operations per Second): How fast do they need to read/write data?

  • Throughput: How much data moves per second (MB/s or GB/s)?

• Network Bandwidth

  • How fast does the server need to send/receive data?

  • 1GbE is basic; 10GbE, 25GbE, or more for large workloads.

• High Availability (HA) and Disaster Recovery (DR)

  • If one server fails, is another ready to take over?

  • DR means data and apps can be recovered if the whole site goes down.

• Virtualization Support

  • Are they using VMware, Hyper-V, or KVM?

  • This affects hardware compatibility and sizing.

1.2 Stakeholder Identification

When planning a server solution, who you talk to matters. Not everyone has the same goals.

You need to talk to all the key people (called stakeholders):

• IT Administrators

  • Know the current infrastructure and technical requirements.

  • Will be responsible for managing the servers.

• Application Owners

  • Understand how their apps use resources (CPU, memory, I/O).

  • Will define how critical performance and uptime are.

• Finance/Procurement Teams

  • Handle budgeting and purchasing.

  • Must understand the return on investment (ROI).

• End Users

  • May not be technical, but their experience matters.

  • Example: “The system is slow” may point to IOPS or CPU issues.

As a designer, your job is to balance the needs of all these people.

1.3 Constraints Identification

Constraints are things that limit your options.

Environmental Constraints

• Rack space: Not all solutions fit in all racks.

• Power: Does the building or data center have enough power?

• Cooling: Do they have proper airflow and HVAC?

Financial Constraints

• Budget Limitations: May force you to find the most cost-effective solution.

• TCO (Total Cost of Ownership): What will the server cost over its lifetime (energy, support)?

• ROI: Will the investment pay off by saving time, increasing productivity, or reducing errors?

Compliance/Security Constraints

• Data residency: Must data be stored within a certain country?

• Encryption: Is hardware or software encryption required?

• Auditability: Are logs needed for compliance reviews?

2. Solution Sizing and Performance Planning

Once you’ve gathered the customer’s requirements, the next step is to figure out how big and powerful the solution needs to be. This process is called sizing — and it includes performance planning, redundancy, and making sure the system can grow.

2.1 Workload Profiling

Before you choose any servers or storage, you must understand what kinds of workloads the customer runs. Each type of workload behaves differently.

Typical Workloads

• Web Services

  • Lightweight but sensitive to latency.

  • Needs stable networking and low CPU overhead.

• DBMS (Database Management Systems)

  • Heavy on I/O and memory.

  • Requires fast storage and sometimes HA clustering.

• File/Print Servers

  • High storage needs.

  • Moderate CPU and memory.

• Virtualization (VMware, Hyper-V, KVM)

  • Multiple virtual machines (VMs) share physical resources.

  • You must size for total combined needs: CPU, RAM, storage, bandwidth.

• AI/ML and HPC (High Performance Computing)

  • Require GPUs or many CPU cores.

  • Need high memory bandwidth and parallel compute power.

Performance Metrics

To size a system properly, track these four key metrics:

• CPU Utilization (%)

  • How heavily the processor is used.

  • Important for apps that do calculations or run scripts.

• Memory Consumption (GB)

  • How much RAM the workload needs at peak usage.

  • Avoid oversubscription to prevent crashes or slowness.

• Storage IOPS

  • IOPS = Input/Output Operations Per Second.

  • Crucial for databases and real-time applications.

• Storage Latency and Throughput

  • Latency = time delay in storage responses (measured in ms or µs).

  • Throughput = volume of data transferred over time (MB/s or GB/s).

• Network Throughput

  • Amount of data flowing through the NICs.

  • High-throughput apps need 10GbE or higher.

2.2 Tools for Sizing

HPE provides special tools to help you size solutions accurately.

HPE Sizer Tools

• HPE ProLiant Sizer for VMware vSphere

  • Helps plan ProLiant server configurations for virtualized environments.

  • You enter the number of VMs, their workloads, and the tool recommends CPU, memory, storage, and networking.

• HPE Right Mix Advisor

  • Helps choose between on-prem and cloud workloads.

  • Evaluates performance, cost, and risk.

Capacity Planning Models

These are simple models or benchmarks to help estimate performance needs:

• Rule of Thumb Sizing

  • Uses standard estimates like:

    • 4 vCPUs per app server

    • 8GB RAM per database

  • Good for quick estimates, but not precise.

• Benchmark Standards

  • SPEC CPU: Measures processor performance.

  • PassMark: Common online CPU and memory benchmarks.

  • TPC (Transaction Processing Council): For databases and transactional systems.

These benchmarks help compare different hardware options.

2.3 Redundancy and Fault Tolerance

Your design should be resilient. That means it must keep working even if part of it fails. Here's how to plan for that.

Redundancy Levels

• N+1

  • One additional component (power supply, fan, server) is available as a backup.

  • Example: If you need 4 servers to run the system, have 5 (1 is a spare).

• N+N

  • Fully mirrored systems — if one half fails, the other keeps working.

  • More expensive, but more reliable.

• Dual Power Supplies

  • Each server has two power inputs.

  • Connected to different power sources (PDU A and PDU B).

• Dual Network Paths

  • Each server has two NICs connected to different switches.

  • Ensures network availability even if one switch fails.

RAID Configurations (Storage Redundancy)

• RAID 1 – Mirroring

  • Data is duplicated on two drives. Safe but halves capacity.

• RAID 5 – Striping with parity

  • Balances capacity and redundancy. Needs at least 3 disks.

• RAID 6 – Double parity

  • Survives two disk failures. Good for large arrays.

• RAID 10 – Mirror + Stripe

  • Best performance and redundancy, but uses more disks.

Clustering & Failover

These ensure services continue running even if one server fails.

• HPE Serviceguard

  • Provides automated failover for Linux workloads.

• VMware HA (High Availability)

  • If one host fails, VMs automatically restart on another.

• Windows Failover Clustering

  • Used for Microsoft SQL Server, File Server clusters, etc.

3. Server Architecture Design

Once you've identified the requirements and planned the system size, it’s time to design the architecture — this means choosing the right servers, storage, and network configurations. This section helps you build a system that performs well, is scalable, and easy to maintain.

3.1 Server Selection Criteria

Choosing the right server form factor and features is critical for meeting the needs of your design.

Form Factor: Rack vs. Tower vs. Blade

• Rack Servers

  • Example: HPE ProLiant DL380

  • Fit in standard data center racks (usually 1U or 2U).

  • Best for general-purpose computing and virtualization.

  • Great for medium to large deployments.

• Tower Servers

  • Example: HPE ProLiant ML350

  • Look like desktop towers.

  • Quiet and easy to deploy in office environments.

  • Ideal for small offices or branch locations.

• Blade Servers

  • Example: HPE BladeSystem

  • Installed into enclosures like HPE c7000.

  • Space-saving and power-efficient.

  • Ideal for data centers with high-density requirements.

CPU Models and Socket Count

• Choose based on performance needs:

  • Intel Xeon or AMD EPYC processors.

  • Some servers support dual-socket or quad-socket CPUs for more cores.

• Higher core count = better for virtualization or parallel processing tasks.

Memory Capacity

• Calculate based on:

  • Number of virtual machines.

  • Database size.

  • Application memory usage.

• Some servers support NVDIMM or NVDIMM-P:

  • Non-Volatile DIMMs — they keep data even after power loss.

  • Great for databases and mission-critical apps.

PCIe Expansion

• Allows you to add:

  • GPUs (for AI/ML)

  • Network cards (1/10/25/40/100GbE)

  • RAID controllers

  • Fibre Channel HBAs

• Important for customizing the server to match application needs.

Hot-Plug vs. Cold-Plug Support

• Hot-plug components (disks, PSUs, fans) can be replaced without shutting down the server.

• Very useful in environments where uptime is important.

3.2 Storage Design

A good design includes enough storage capacity, performance, and redundancy.

Local Storage (inside the server)

• SAS: Enterprise-grade, reliable, good for performance.

• SATA: Lower cost, good for archive or low-IO workloads.

• NVMe: Extremely fast; used for caching or high-performance apps.

• M.2 drives: Small form factor SSDs often used for boot drives.

Shared Storage (external)

• SAN (Storage Area Network):

  • High-performance block-level storage.

  • Uses Fibre Channel or iSCSI.

• NAS (Network Attached Storage):

  • File-level storage accessed over TCP/IP.

  • Simpler to manage but usually slower than SAN.

• HPE Primera and Nimble:

  • HPE’s enterprise-grade storage arrays.

  • Offer high availability, InfoSight integration, deduplication, and more.

Software-Defined Storage

• vSAN (by VMware):

  • Uses local disks to create shared storage in a VMware cluster.

  • Cost-effective alternative to SAN.

• HPE SimpliVity storage:

  • Built into HPE hyperconverged systems.

  • Includes backup and deduplication features.

3.3 Network Design

Servers need strong, fast, and redundant network connections — especially for virtualization and storage access.

Interface Types

• 1GbE: Basic standard; fine for management or light workloads.

• 10GbE: Common in virtualization and storage networks.

• 25/40/100GbE: Used in high-performance computing, large-scale data transfers.

Redundancy Models

• NIC Teaming / Bonding:

  • Combines two or more network interfaces for:

    • Increased bandwidth

    • Failover protection

• LACP (Link Aggregation Control Protocol):

  • Industry standard method of link bonding.

• Redundant Uplinks:

  • Connect servers to two separate switches to avoid single points of failure.

HPE Networking Options

• Virtual Connect:

  • A module in HPE BladeSystem or Synergy.

  • Simplifies network management by abstracting physical connections.

• FlexFabric Modules:

  • Provide high-speed, converged networking (LAN + SAN).

  • Used in Synergy to manage compute and storage connectivity in one place.

4. Virtualization and Cloud Readiness

Modern IT environments rely heavily on virtualization (running multiple virtual machines on a single physical server) and cloud services. Your server design must fully support these technologies — both for current use and future scalability.

4.1 Hypervisor Integration

A hypervisor is software that runs virtual machines (VMs) on a physical server. HPE servers must be compatible and optimized for the hypervisors your customer is using.

Supported Hypervisors:

• VMware ESXi

  • Industry standard for enterprise virtualization.

  • ProLiant servers are certified and optimized for VMware.

  • Supports vCenter, vMotion, vSAN, DRS, HA.

• Microsoft Hyper-V

  • Integrated into Windows Server.

  • Great for Windows-centric environments.

  • Supports clustering, live migration, and SCVMM integration.

• Linux KVM

  • Open-source, cost-effective.

  • Common in cloud and DevOps environments.

• Nutanix (on HPE ProLiant DX Series)

  • Hyperconverged platform combining compute, storage, and virtualization.

  • Uses AHV (Acropolis Hypervisor) or VMware.

  • Preloaded on certified ProLiant DX servers.

Design Tip: Always check the hypervisor compatibility list for the chosen HPE server model (via VMware HCL, Microsoft HCL, etc.).

4.2 Virtualization Features Planning

When designing for a virtualized environment, there are specific features and requirements you must plan for:

vMotion / Live Migration

• What it is: Move a VM from one physical host to another without downtime.

• Why it's important: Ensures availability during maintenance or failure.

vSphere DRS (Distributed Resource Scheduler)

• Automatically moves VMs between hosts for:

  • Load balancing

  • Energy savings (power off unused hosts)

High Availability (HA)

• If one host fails, affected VMs are automatically restarted on another host.

• Requires shared storage or a hyperconverged setup like SimpliVity or vSAN.

Virtual Networking Design

• Plan for:

  • vSwitch or Distributed vSwitch setup

  • VLAN tagging and isolation

  • NIC teaming (for redundancy)

Storage vMotion and vVols

• Storage vMotion: Move a VM’s disk to a different datastore without downtime.

• vVols (Virtual Volumes):

  • HPE storage (e.g., Nimble, Primera) supports vVols.

  • Allows per-VM storage management, better snapshots and backups.

4.3 Hybrid Cloud Design with HPE

A hybrid cloud uses a mix of on-premises infrastructure and public cloud. HPE offers several tools to help customers integrate with the cloud.

HPE GreenLake

• Cloud experience delivered on-prem.

• Pay-per-use pricing.

• Scales like the cloud but runs in the customer’s data center.

• Supports workloads like SAP HANA, VMware, VDI, AI/ML, etc.

HPE Cloud Volumes

• Cloud-based block storage.

• Replicate or back up data to the cloud (AWS, Azure, GCP).

• Easily move volumes between cloud and on-prem.

HPE Cloud Data Services

• Backup as a Service: Automatically backup virtual machines and files to the cloud.

• Disaster Recovery as a Service: Spin up systems in the cloud if your site goes down.

Design Tip: When planning for hybrid cloud, you must consider:

• Security and compliance for offsite data.

• Bandwidth and latency.

• Integration with identity and access control (e.g., Active Directory, IAM).

5. Management and Security Design

A well-designed server solution isn’t just about performance — it must be secure, easy to manage, and ready for automation. This section focuses on tools and technologies that help manage infrastructure and protect it from threats.

5.1 Infrastructure Management

Efficient server management is key to reducing manual work, increasing uptime, and ensuring firmware compliance. HPE provides powerful tools to manage servers from setup to retirement.

HPE OneView

What is it?

• A centralized platform for managing servers, storage, and networking.

Key Features:

• Role-Based Access Control (RBAC)

  • Assign roles like admin, operator, viewer.

  • Controls what users can see or change.

• Template-Based Provisioning

  • Create Server Profiles that include:

    • BIOS settings

    • Network/storage connections

    • Firmware baseline

  • Apply the profile to any server to configure it automatically.

  • Reduces errors and speeds up deployment.

• Firmware Compliance

  • Monitor and enforce that all servers meet the same firmware standard.

  • Detects outdated or inconsistent versions.

• Monitoring and Alerts

  • Get real-time system health updates.

  • Integrates with tools like SNMP, email alerts, or SIEM systems.

iLO 5 (Integrated Lights-Out)

What is it?

• A dedicated management chip built into HPE ProLiant servers.

Key Capabilities:

• Remote Console

  • View and control the server remotely (like a virtual keyboard/mouse).

  • Access even if the OS isn’t running.

• Secure Boot

  • Verifies firmware integrity during power-up.

  • Prevents boot if the firmware is tampered with.

• Hardware Health Monitoring

  • Watch CPU temperature, memory errors, fan status, power draw.

• Intelligent Power Management

  • Monitor and optimize power usage.

  • Helps with cooling planning and cost control.

• Two-Factor Authentication and Directory Integration

  • Works with LDAP/AD, and supports 2FA for added security.

5.2 Security Considerations

Security is a major concern for all IT systems. HPE servers offer built-in security features that help protect data, firmware, and user access.

Silicon Root of Trust

• A hardware-level fingerprint that verifies the integrity of server firmware at boot time.

• If tampered firmware is detected, the system halts the boot process.

• Prevents firmware-based attacks (e.g., ransomware injected into BIOS).

Secure Erase of Storage Devices

• When a server is decommissioned or repurposed, you can perform a secure erase of drives.

• Ensures that no residual data is left on HDDs, SSDs, or NVMe drives.

• Complies with data protection regulations (e.g., GDPR).

iLO User Management and 2FA

• Use directory-based accounts (e.g., Active Directory) or local iLO users.

• Set password policies and expiration rules.

• Enable two-factor authentication (2FA) for extra protection.

Firmware Validation and Rollback

• iLO and OneView can validate firmware signatures.

• If an update fails or causes issues:

  • Rollback to the last known good firmware version.

  • Avoids server downtime due to bad updates.

Best Practices Summary:

Area	Best Practice
Access Control	Use RBAC and directory integration (LDAP/AD)
Patch Management	Maintain firmware baseline across all servers
Physical Security	Rack locks, secure rooms
Network Security	Use separate VLANs for iLO and production traffic
Data Protection	Use secure erase before disposal or reuse of drives

6. Documentation and BOM (Bill of Materials) Preparation

Even the best-designed server solution is incomplete without documentation. This step ensures that everything you've planned can be clearly communicated to technical teams, procurement departments, and support engineers.

6.1 Technical Architecture Diagram

This is a visual blueprint of your solution. It should include both logical and physical elements.

Logical Layout

• Shows how components interact.

• Include:

  • Virtual machines and their roles (e.g., DB, web, app)

  • Network segmentation (management, production, storage)

  • Storage architecture (vSAN, SAN, NAS)

Physical Layout

• Shows actual server and cable placement in racks.

• Include:

  • Server names and models

  • IP addresses

  • Cable runs (to switches, storage, power)

  • Rack elevation (server position in rack)

  • Numbering for PDU connections

A well-documented layout helps:

• Avoid cabling errors

• Speed up deployment

• Support troubleshooting

6.2 BOM (Bill of Materials) Details

A BOM lists everything needed to build and deploy your solution. This goes to procurement teams and ensures the project is fully scoped.

Typical BOM Includes:

• Server Models and SKUs

  • e.g., HPE ProLiant DL380 Gen10, Part #P12345-B21

• Add-ons

  • CPUs (exact models and quantity)

  • RAM (type, speed, quantity)

  • Storage (type: SSD/HDD, interface: SAS/SATA/NVMe)

  • Network cards or GPUs

  • RAID controllers

  • Power supplies (standard or redundant)

  • Rail kits and mounting kits

• Software Licenses

  • VMware, Windows Server, HPE OneView Advanced, Veeam, etc.

• Management & Support

  • iLO Advanced license

  • HPE Insight Remote Support

  • HPE InfoSight integration (for supported platforms)

• Services

  • HPE Pointnext (installation, setup, and lifecycle services)

  • On-site support, 4-hour or next-business-day SLA

Tip: Every item in the BOM must include a part number (SKU) and description, so the procurement team can easily place orders.

6.3 Lifecycle Cost Estimate

A good design should also show the cost across the entire server lifecycle, not just the purchase price.

CAPEX (Capital Expenditures)

• Upfront cost of:

  • Hardware

  • Licenses

  • Installation and setup

OPEX (Operating Expenditures)

• Ongoing costs like:

  • Electricity and cooling

  • Support contracts

  • Staff to manage servers

  • Licensing renewals

Licensing Considerations

• Example: VMware often licenses per CPU socket, so dual-CPU servers may cost more in licenses.

• Microsoft Windows Server may be licensed per core.

• Be aware of OEM licenses, subscription models, and bundled support.

Summary

Deliverable	Purpose
Architecture Diagrams	Help IT and cabling teams understand how everything fits
BOM List	Helps procurement buy exactly the right parts
Cost Estimate	Helps decision-makers understand upfront and long-term costs
Security & Management Plan	Ensures the system is secure, compliant, and easy to run

Plan and Design Server Solutions (Additional Content)

Common Sizing and Design Mistakes to Avoid

Even with the best planning tools, poor assumptions or oversight can lead to costly performance issues, downtime, or underutilized infrastructure. Below is a breakdown of frequent design errors, the impact, and how to avoid them.

1. Ignoring Power and Cooling Redundancy

Mistake:

Design does not include redundant power supplies or fails to consider rack-level cooling limits.

Symptoms:

• Unexpected shutdowns

• Component damage due to thermal stress

• Limited uptime during utility or HVAC failures

Impact:

High risk of system failure during power interruptions or cooling degradation — especially in high-density deployments or hot climates.

Best Practices:

• Always design with dual PSUs, each connected to separate PDUs or UPS circuits (PDU A / B)

• Use iLO or OneView to monitor power draw and thermal margins

• Validate rack airflow direction (front-to-back) and avoid mixing airflow patterns

• Factor in ASHRAE environmental standards (18°C–27°C temp, 40–60% humidity)

2. Overcommitting Virtualization Resources

Mistake:

Sizing based on total vCPU or memory requirement, without considering oversubscription limits.

Symptoms:

• Slow VM performance

• Contention for CPU or RAM

• Host crashes under peak load

Impact:

Performance bottlenecks in production environments, especially under load surges or during VM migrations.

Best Practices:

• Use realistic consolidation ratios (e.g., 4:1 vCPU to pCPU for general workloads, lower for DB or high IOPS systems)

• Plan for headroom (20–30%) to accommodate spikes or failover

• Avoid memory oversubscription unless using memory ballooning or swapping features wisely

• Use HPE ProLiant Sizer for VMware vSphere to validate VM density and host capacity

3. Underestimating Network Interface Requirements

Mistake:

Designing with too few NICs or ports, assuming minimal network throughput.

Symptoms:

• Backup jobs fail or run slowly

• vMotion or iSCSI traffic congests production VLANs

• No redundancy in case of NIC or switch failure

Impact:

Traffic collisions, latency spikes, and single points of failure for key workloads (e.g., storage, backup, live migration)

Best Practices:

• Use dedicated NICs or vNICs (via FlexFabric or Virtual Connect) for:

  • Management

  • Backup

  • VM traffic

  • Storage (iSCSI/NFS/Fibre Channel over Ethernet)

• Implement NIC teaming (Windows) or bonding (Linux) for redundancy and load balancing

• Provision minimum 10GbE uplinks for virtualization clusters or storage-heavy apps

4. Ignoring Future Scalability

Mistake:

Sizing only for current needs, without capacity for growth or change.

Symptoms:

• Need to replace hardware after a short period

• Project delays due to procurement or datacenter constraints

Impact:

Wasted investment and higher TCO due to repeated upgrades or premature obsolescence.

Best Practices:

• Include a growth buffer (20–30%) for CPU, RAM, disk, and NIC usage

• Use modular platforms (e.g., HPE Synergy, DL380 Gen11 with PCIe expansion) to allow scaling without forklift upgrades

• Document scaling limits per rack, chassis, or platform

5. Misconfigured Storage IOPS or RAID Levels

Mistake:

Selecting storage based on capacity only, not performance.

Symptoms:

• Slow database queries

• High latency for VMs or apps

• Long RAID rebuild times

Impact:

I/O bottlenecks that hurt critical workloads, especially for databases or virtual desktops

Best Practices:

• Estimate IOPS and latency requirements per workload

• Choose RAID levels wisely:

  • RAID 10 for write-heavy workloads

  • RAID 6 for large-capacity arrays

• Consider NVMe SSDs or HPE Smart Cache for hot data acceleration

• Use HPE SSA or Intelligent Provisioning to configure RAID with appropriate stripe size and cache policies

6. Failing to Validate Compatibility or Interoperability

Mistake:

Choosing hardware or firmware without checking OS, hypervisor, or application compatibility.

Symptoms:

• Unsupported drivers

• Firmware-induced crashes

• Failed OS installs

Impact:

Project delays, support calls, and possible security exposure

Best Practices:

• Check HPE Server QuickSpecs and VMware/Microsoft HCLs

• Use SPP (Service Pack for ProLiant) for validated firmware bundles

• Always test with a pilot system before scaling deployment

Summary: Design Pitfalls Checklist

Pitfall	Preventive Action
Power/cooling ignored	Dual PSUs, thermal design, iLO sensor checks
Virtualization overcommitment	Realistic VM-to-host ratio, capacity headroom
Too few NICs	Separate traffic types, enable teaming/bonding
No growth margin	Design for 12–24 months expansion
Storage design flawed	Match IOPS to RAID type, use SSD where needed
Compatibility issues	Cross-check firmware/OS/hypervisor matrix