1U rack server full range of features and management analysis

Today, to meet the growing demand for IT, enterprises and organizations will buy many servers. But why did you decide to use a 1U server? The answer is simple. These servers are a fundamental and important part of the data center—especially for small and medium-sized businesses to build their own IT infrastructure. These low-cost, powerful and space-saving "small" servers hit the key to small and medium-sized businesses with tight IT budgets, and they don't have to worry about performance sacrifices. The 1U server is fully capable of tasks such as the day-to-day business automation engine and other application operations. You can use a 1U server to run a website, deploy a file service or print service, and support network tasks. However, these rack-mounted servers are not suitable for running database or enterprise-level massive information processing that requires extremely high computational performance. The 1U server can even meet the virtualized application environment and can run multiple virtual machines (VMs) at the same time to enhance the utilization value of the device.

Of course, the performance of different servers is naturally different, and there are big differences in server models, functions, and computing resources (for example, CPU, memory, and I/O). As a server buyer, you need to be able to select a server that fits your enterprise application among a wide variety of brands and models. As virtualization becomes more widely used, buyers will pay more attention to the virtualization features of servers.

This article will introduce the key factors needed to understand the enterprise 1U rack server, compare the internal features, expansion capabilities and management functions of various 1U servers, and provide a current mainstream for buyers. A list of specifications for the 1U server.

1U Server Internal Features

Processor. Of the many feature-rich 1U servers, processors are often the primary consideration. Today's 1U servers can typically support 1 or 2 CPUs, such as the Intel Xeon 5500 or Xeon 5600 series. AMD processor-based servers typically use the Opteron 4100 or Opteron 6100 series. The Fujitsu Primergy RX200 S6 1U server (see Table 1) uses 1 or 2 Xeon 5600 series processors, while Dell's PowerEdge R415 1U server uses 1 or 2 Opteron 4100 series processors.

There are many models to choose from for each processor family; each processor has a unique mixed clock rate, number of cores and cache size. Current processors typically contain 6 to 8 cores. When we multiply the number of processors by the number of cores and optimize them by processing instructions (such as Intel Hyper-Threading (HT)), each 1U server can provide amazing computing power. As long as the processor resources and computing power are managed through a hypervisor, multiple virtual machines (VMs) can be run on the server. You can also select only one processor when purchasing a server, and then perform hardware upgrades later. Keep in mind that in general, the processor needs to be consistently matched to ensure that the model number and speed are the same as the manufacturing steps (revision or version).

Memory. The second important factor to consider is server memory. Virtual machines typically run out of memory (or RAM) before they run out of processor performance. Organizations deploying server virtualization should pay more attention to memory rather than processors. Today's 1U servers can support a lot of memory. Take Super Micro Computer's 1026T-RFT+ 1U server as an example, you can manage 18-slot dual in-line memory modules (DIMMs) as small as 2GB and as large as 192GB. The Dell PowerEdge R415 1U server supports up to 128GB of DIMM slots.

It is worth noting that such large-capacity memory typically uses registered (cached) DIMMs. It also supports non-cached DIMMs, but the total amount of memory supported is much lower -- perhaps only half of the cached DIMMs Or one third. Note that memory is usually allocated to each processor, so a 1U server with 2 processors and 12 DIMM slots can manage 6 DIMM slots per processor. Almost all 1U servers use the current fastest DDR3 memory to match the current processor bus (front end) speed.

There are some features in memory that deserve attention. The Error Correction Code (ECC) is the current mainstream memory and can correct single bit errors in memory and detect double bit errors. Server administrators can also choose non-ECC memory, but give up detection and error correction, which can cause critical application failures. The 1U server running mission-critical tasks also benefits from single-device data correction (SDDC) memory, which extends the ECC mode to check and correct memory chip-level errors.

The functions and principles of memory hot backup and disk hot backup are the same. Take the extended memory module as an example, it has been installed, but it is not used. Once a memory device fails, hot spare memory will replace the failed memory until the failed memory is replaced and returned to normal. Memory mirroring is similar to RAID 1 (disk mirroring). The data is copied to two different memory regions and compared. If an uncorrectable error occurs in a memory area, the system will get the data from the mirrored area.

Storage. Storage is another feature of the 1U server. Today's products can support a variety of miniaturized modules (usually 2.5-inch) SATA (Serial ATA) and SAS (Serial Attached SCSI) disks, usually with an optical drive, such as A small CD-ROM or DVD-ROM is used to install the software. How to deploy a server and how much local storage is needed is a question that deserves careful consideration. If the server is running a load on a local disk, it needs to provide sufficient capacity and performance.

Taking SAS as an example, SAS disks offer low capacity and high performance, making them ideal for general applications. SATA offers higher capacity and lower performance, making it ideal for file and print servers. The HP Proliant DL360 G7 and Dell PowerEdge R415 1U servers also support Solid State Drives (SSDs). Although the current price of solid-state drives is extremely high, it can provide excellent services for storage-sensitive applications such as databases and Exchange Server 2010. The number of disk bays depends on the server model and options, and some 1U servers do not have a disk installed by default. You need to check that the server unit configuration meets your needs before placing an order. Servers that plan to use shared storage do not have to worry about how the local disk should be configured.

When administrators use local disks, they need to consider RAID technology to protect data. Modern 1U servers support hardware-based RAID such as RAID 0, RAID 1, RAID 5, and RAID 6, as well as RAID 10, RAID 50, and RAID 60 modes. If the disk controller located on the server board cannot support RAID, you will need to add a separate RAID controller (with backup battery) to the expansion slot of the server.

Expansion slot. A 1U server typically provides some scalability through one or more slots. These slots are usually designed to install expansion boards on the motherboard. You can install enhancement boards such as RAID controllers, network adapters, Fibre Channel host bus adapters, etc. to connect to storage area networks (SANs) or other adapters. hardware. Need to pay attention to distinguish between PCIe (PCI Express) and PCI-X (PCI-eXtended) slots - two types of boards are not interchangeable.

Taking the Super Micro 1026T-6RFT+ 1U server as an example, the server supports a thin PCIe x4 (4-channel) board and can support two PCIe x8 (8-channel) boards. In contrast, the HP Proliant DL360 G7 supports one PCIe x8 and one PCIe x16 board, but if the riser card is replaced, the server can support PCIe or PCI-X cards. You need to ensure that the upgrade card for the server matches the server's internal space and the number of slots in the slot. Otherwise it may not be available.

Power consumption. Each server needs to transmit energy over a standard alternating current (AC) line, requiring output voltage and type matching cables, which may vary from region to country. Power problems are worthy of attention, with a focus on server power supply auto-tuning and fault tolerance.

The Super Micro 1026T-6RFT+ 1U server provides redundant, hot-swappable 700W power modules—each power module at the rear of the chassis supports independent AC power. Other servers, such as the HP Proliant DL360 G7, have a single 460W power supply by default, and can also support additional redundant power supplies or use a higher power 750W power supply. If multiple CPUs, large amounts of memory, extended drives, and some expansion boards are configured, the server needs a higher energy supply.

Redundant configuration has great advantages. If a single power supply fails, the server will still work. Not only that, but the server can achieve good fault tolerance through two independent AC power supplies. The server can also be powered by two separate uninterruptible power supply (UPS) systems. Organizations with tight budgets prefer to use one power supply to connect to the power strip and another to connect to the UPS. This saves the purchase of another UPS cost as a redundant power supply.

Network connectivity. More and more administrators are considering configuring more network adapters for the server. The LAN interface is one of the key components of modern servers, and the number of required network cards needs to be determined according to actual needs. Ordinary commercial servers usually have only one Gigabit Ethernet port, and more advanced servers usually provide multiple Gigabit interfaces.

If you plan to deploy server virtualization and run multiple virtual machines on a single server, you should choose a server with multiple network controllers. If you have high-end network connection requirements (such as the server needs to handle streaming media or massive user load), consider adding 1~2 10 Gigabit network controllers to the 1U server. For example, the Super Micro 1026T-6RFT+ 1U server can provide two Gigabit and two 10G interfaces at the same time.

The system management feature is the function provided by various types of servers. The Intelligent Platform Management Interface (IPMI) is the core of server management. The IPMI-enabled server includes a dedicated low-bandwidth LAN interface that sends Simple Network Management Protocol (SNMP) alert information (and other control information) over the network. Administrators can use IPMI traffic to check server status, hardware logs, or perform other administrative tasks. The server may also have a traditional serial port (RS-232), but with the popularity of network communication, the serial port is rarely used.

User Interface Port. The server also provides a variety of user interface ports for connecting keyboards, monitors, and mice (KVM), as well as multiple USB ports.

Typically, by connecting a user interface port to a central KVM switch, administrators can manage the server with a single mouse, keyboard, monitor, and mouse.

The current popular solution is to use KVM-over-LAN or KVM-over-IP technology to manage servers through remote management consoles. IP-based remote management supports LAN and WAN.

The USB interface can be connected to a USB keyboard, mouse or storage medium (for installing software), dongle and other peripherals.

Panels and indicators. Many IT people usually don't pay attention to the control panel and indicators located on the front of the server. They believe that diagnostic or system management tools can already provide sufficient status and fault information, but one should not underestimate the role of the server status indicator.

Front panel indicators -- such as power, hard drive status, network activity, temperature status alarms, power and fan failures -- can diagnose or provide early warning of potential problems without having to bother Traditional diagnosis. Some servers include a series of indicators, while other systems, such as the HP ProLiant DL360 G7, provide a "health" indicator that reports basic faults in different ways.

An indicator located on the server's stand-alone drive bay displays the health of each individual disk and identifies the failed physical disk for replacement. Also consider the availability of the drive bay. Almost all current 1U servers offer hot-swappable hard drive bays. For example, the Super Micro 1026T-6RFT+ 1U server (see Table 1) provides eight hot-swappable bays for 2.5-inch SAS and SATA hard drives. The HP Proliant DL360 G7 provides four hot-swap bays by default for installing four expansion hard drives.

Each drive is mounted on a pallet for easy installation and removal. Unlike traditional computers that require a cable to be connected to the system board, the 1U server provides a fixed backplane that allows the drive to be mounted to the server along with the bay and locked to the chassis.

Tip: If you are sourcing a server for deploying shared storage (such as server virtualization), you should have almost no need to configure a local hard drive. Instead, the server will boot through a network share and load and run on the SAN. virtual machine. The server can also support Fibre Channel SAN by adding a Fibre Channel interface card to the slot. If you plan to use an Ethernet-based SAN (such as iSCSI), you only need to connect through one of the server's LAN interfaces.