DC Details: Battle of the Blades

In the year 2000, RLX Technologies released the blade server platform to the world. Since then, Dell, HP Enterprise, Cisco, Lenovo and many others have released their own variants. While they never usurped rack servers, their compact, modular design are an ideal choice for industries requiring high density computing.

Blades are used particularly by cloud service providers, as they enable them to maximize server density in data centers, offering scalable and efficient resource management to meet varying customer demands. They also see usage in financial, government and healthcare industries. Let's take a detailed look at the technologies involved and then compare two key manufacturer's offerings.

Blade Breakdown

A Blade system consists of a chassis, servers, management modules, network modules, storage modules, power supplies, cooling and a backplane.

Blade Chassis

The chassis or enclosure, is the physical housing that contains and powers the blade servers. It provides shared resources like power, cooling, networking, and management.
Key Features include:

• Power Supplies: Delivers power to all the blades in the chassis.
• Cooling: Integrated fans and airflow systems cool the blade servers.
• Management Modules: Provide centralized control and management of all blades within the chassis.

Blade Servers

The servers are the individual computing units that slide into the chassis. Each blade is a full server, typically including its own CPU, memory, storage, and network connectivity.
Key Features include:

• Processor (CPU): Each blade has its own processors, typically multi-core, providing the compute power.
• Memory (RAM): Upgradeable modules, installed directly on the blade for running applications and processes.
• Storage: Some blades may have local storage (e.g., SSDs or HDDs), while others might depend on shared storage solutions.
• Network Interfaces: Network connectivity is provided through adapters or mezzanine cards, connecting to the chassis’s shared networking components.

Management Modules

The management modules (management controllers or chassis management controllers) provide centralized management, monitoring, and configuration of all the blade servers and other components within the chassis.
Key Features include:

• Remote Management: Allowing administrators to remotely manage blade servers, including power cycling, firmware updates, and system health monitoring.
• User Interfaces: Typically provides a web-based or command-line interface for managing the entire blade system.
• Integrations: Often integrates with broader data center management tools.

Network Modules

These modules (also referred to as switches and fabric modules) provide network connectivity for the blade servers, aggregating and routing traffic to external networks and other data center resources.
Key Features include:

• Ethernet Switches: Provide Ethernet connectivity for the blade servers, often supporting high-speed connections (e.g., 10GbE, 25GbE).
• Fiber Channel Switches: Used for connecting to storage area networks (SANs).
• Fabric Modules: These handle more complex network fabric connections, integrating various network types (Ethernet, Fibre Channel, etc.).

Storage Modules

These are shared storage modules, such as storage blades or integrated storage arrays, accessible by all blades within the chassis.
Key Features include:

• Direct-Attached Storage (DAS): Some systems include storage blades with dedicated storage that can be directly attached to specific blades.
• Network-Attached Storage (NAS) or SAN: This involves an integration with external storage networks for high-capacity storage solutions.

Power Supplies

The power supply units (PSUs) deliver power to the entire chassis and all the blades within it.
Key Features include:

• Efficiency: High-efficiency PSUs are used to minimize energy consumption.
• Redundancy: Multiple PSUs provide failover protection and redundancy.
• Hot-Swappable: Can often be replaced without shutting down the chassis.

Cooling Modules

Like any other hardware, the blade system produces heat. Cooling modules, typically in the form of fans or liquid cooling systems, maintain optimal operating temperatures for all the components within the chassis.
Key Features include:

• Variable Speed: Fans adjust their speed based on temperature sensors within the chassis.
• Redundancy: Multiple fans provide redundancy to prevent overheating.

Interconnects/Backplane

The backplane is an internal circuit board within the chassis that connects the blade servers to shared resources like power, networking, and storage.
Key Features include:

• High-Speed Data Transfer: Facilitates high-speed data communication between blades and other components.
• Scalability: Supports the addition or removal of blades without affecting the overall system.

The blade system is a nicely packaged solution for providing servers in your datacenter, with all the features needed for a modern solution. As mentioned Dell and HPE are two manufacturers of blades.

Dell Blades

I've worked with these blades first hand. I'll do my best to provide an unbiased review.
Dell Introduced their PowerEdge blade server in 2005, since then it has gone through several generations. The PowerEdge MX-Series are their more recent offering, providing for increased flexibility, scalability, and support for emerging technologies like NVMe and high-speed networking fabrics.

Where they Shine

Dell's proprietary iDRAC (integrated Dell remote access controller) is a tool for managing deployments, updates and monitoring of PowerEdge blades from anywhere. It offers a secure solution using TLS encryption, smart card authentication, LDAP, automated certificate enrollment, 2-factor authentication and a system lockdown mode. In the automation department, iDRAC enables APIs such as Redfish as well as orchestration tools like Ansible and Racadm. This permits automated management of thousands of servers. They also provide telemetry data, providing analytics data on the server health and allowing you to make proactive decisions to optimize or reduce downtime. It's a great tool.

Real customers have claimed that, "The initial setup is straightforward." They've also claimed, "The installation is easy." Swapping blade servers in the chassis in and out is easy

Firmware updates to PowerEdge servers are efficient and don't require downtime for individual blades.

Where they're Dim

There have been reports of issues with RAID configurations, particularly with the management software GUIs and driver selection.

Dell offers a unified approach to interconnections, however its been reported to have fewer options than the competition.

As for pricing, PowerEdge systems are more expensive.

HPE Blades

HP introduced their proliant blade server in 2002. Since that time, their solutions have gone through a number of phases, with a more recent offering being their Gen10 Proliant series. They offer both a more traditional rack mounted solution and modular solutions via their Apollo shared infrastructure chassis.

Where they Shine

HP ProLiant servers are known to be reliable. They undergo rigorous testing and quality control to ensure they can handle the demands of mission-critical workloads.

HPE's Silicon Root of Trust is embedded in the HPE Integrated Lights-Out (iLO) management processor, ensuring that the server firmware is securely anchored in the silicon itself. This prevents the server from booting with compromised firmware, offering protection against firmware attacks. Secure boot is also a feature of these servers. HPE ProLiant blades support Secure Erase, which allows administrators to quickly and securely erase data from drives, ensuring that sensitive information is completely removed before decommissioning or repurposing the hardware.

Many HPE ProLiant blades are available with configurations that meet FIPS 140-2 (Federal Information Processing Standard) compliance. This is a desired featured for use cases in the government sector.

The System Lockdown feature allows administrators to lock the server configuration, preventing unauthorized changes to the firmware, hardware, and software configurations. If any changes are attempted, alerts can be triggered.

Proliant series solutions offer diverse interconnect options across different platforms (ProLiant, Synergy, Apollo), with strong support for high-performance networking like InfiniBand, as well as advanced features like persistent memory integration.

Where they're Dim

The Firmware update process has been reported to be more challenging with Proliant blades, "any firmware upgrade with HP will result into issues, this is not the case with Dell servers". "I would definitely like to see them fix the firmware updates on all the blades. No matter what operating system it is running on, make sure I can do the firmware as well as the software for those firmware components at the same time."

In recent reporting by an independent testing firm, Proliant DL servers were shown to handle nearly 3 times fewer VDI instance workloads than MX servers. Proliant servers also reportedly handled 25% fewer orders per minute in a simulated online store test.

Support has been less than dependable at lower tiers. There have been reports of the need to purchase premium support options in order to receive proper issue resolution.

Which is better overall?

Dell's management software and especially their firmware update processes seem to be a better option, with the ability to update the chassis or individual blades. The setup of PowerEdge hardware is also reportedly easier. The number of blades that can be installed per chassis are also typically higher than with ProLiant chassis.

HPE seems to have more thought into enterprise security features. While their firmware update process is reportedly more cumbersome, the integrity of running firmware can be well trusted on the Proliant platform.

All-in-All, PowerEdge seems to have a real edge in terms of real use case workloads, outpacing Proliant servers in nearly every category of testing by an independent firm. This to me is the main differentiator. When it comes down to it you are buying a blade system for its modular design and cloud ready processing capabilities. Any good datacenter management involves placing hardware that is physically efficient as much as it is processing efficient. If I we're recommending a blade solution, I'd pick Dell based on data and real user comments.