Not logged in » Login
May 09 2016

“The simpler the better” – Remote Management with ServerView® eLCM

Today's server systems perform legions of critical functions they weren't tasked with five or ten years ago. As a result, the systems themselves are getting ever more complex, and the same goes for the management duties of a typical server administrator. At the same time, users increasingly depend on reliable IT services and infrastructures, and are becoming less tolerant of delays or downtimes. IT departments are thus forced to look for 'management methods' that will allow them to handle the growing complexity with more ease and confidence. One of these methods is called out-of-band or lights-out management (LOM) and has undergone substantial improvements in the latest edition of Fujitsu's ServerView® Suite.

Out-of-band management uses a dedicated management channel that enables admins to monitor and manage servers via remote control regardless of whether the server is powered on. In Fujitsu's PRIMERGY servers, this capability is implemented via the integrated Remote Management Controller (iRMC S4). The iRMC S4 is an autonomous system that resides on the system board of a PRIMERGY server and is equipped with its own operating system, web server, Linux file system as well as a separate user management and independent alert management. Moreover, it remains powered on even as its host system switches to standby mode.

All of this may sound somewhat boring to experienced PRIMERGY administrators. But even for them, the latest version of the iRMC S4 contains a little surprise package in the form of functional enhancements: the controller now comes with an integrated SD card and allows for comprehensive lifecycle management of a PRIMERGY server from within the system. Since all required functions are largely integrated ("embedded") into and entirely controlled by the iRMC S4, we subsume them under the term "embedded Lifecycle Management," or eLCM for short.

Why Embedded?
Server administrators frequently face numerous challenges that further complicate their already demanding jobs. Oftentimes, these obstacles are relatively small – for instance, handling a batch of installation media like DVDs and USB sticks is definitely inconvenient, but not an insurmountable difficulty. More problematic is the fact that IT infrastructures keep expanding, whereas IT headcount remains comparatively stable (or even diminishes). For server admins, this means they usually don't have physical access to any server to perform hands-on management and maintenance, e.g. to install or upgrade software, check system health, solve RAID issues or find the reason for malfunctions. Under these circumstances, they will frequently find it hard to do their job with the diligence and accuracy it deserves. By contrast, embedding lifecycle management functionality in the firmware of the out-of-band controller enables administrators to control all of those processes with just a few mouse clicks in the iRMC S4 web interface. More specifically, eLCM covers the following areas and functions:

  • Update management
  • Image management
  • PrimeCollect
  • Installation management (eIM)
  • RAID management (eRM)
  • Diagnostics management (eDM)

The first three functionalities have been available since early 2015; eIM, eRM and eDM are to follow in Q2/2016.

Update Management
To ensure 'their' systems have the latest OS, drivers, firmware, and security features installed, server administrators usually compare the latest information from an update catalog at regular intervals. As a time-consuming routine, this task basically lends itself to being automated, and that's what we try to achieve with eLCM update management.

To carry out eLCM updates, the iRMC S4 automatically downloads the required files from a repository server via its dedicated management LAN port and stores them on the built-in SD card. The repository server can be the official Fujitsu support server or a separate system running the ServerView® Repository Server. To further simplify the process, the iRMC can be configured in a way that it will only download sources that fit with the system it resides on. That way, it turns into a partial mirror of the repository server that holds local copies of the matching upgrade catalog, ISO images, and binaries. Some server components can only be updated by the server operating system. To make the corresponding update packages available, the iRMC S4 downloads them, creates corresponding ISO images, and mounts these images as virtual CD-ROM drives.

In this setup, three different update modes are supported: autonomous, online and offline. Each of these may either be initiated automatically (e.g. via timer settings) or manually (via the iRMC S4 web GUI). The individual update modes differ in terms of which server components can be updated and whether the server operating system must be running or not:

  • Autonomous updates will only refresh the iRMC S4 firmware and server BIOS. The iRMC S4 will carry out all necessary operations on its own. Autonomous updates are an integral part of online updates, hence they can't be launched as a separate process; however, they may also be combined with offline updates for a more elaborate process.
  • Online updates extend the autonomous update capabilities, but require that the ServerView Agentless Service runs on the OS of the system in question. This service provides inventory data for firmware and drivers and installs firmware updates while the server is up and running, i.e. during regular operating hours. On Windows servers, it will also install the necessary component drivers.
  • Offline updates are the method of choice if no Agentless Service is running on the managed server or if a server OS is not (yet) supported. As the name indicates, offline updates require that a system or application is allowed a short period of downtime. This method is particularly useful if an admin wants to update the firmware running on components like network or storage controllers.

Image Management
This function mainly serves to download bootable ISO images to the internal SD card. eLCM image management supports two types of images, namely eLCM images provided by Fujitsu or so-called custom images provided by IT departments themselves. Images and image updates may be downloaded from various internal or external sources via FTP, HTTP or their secure counterparts. Once downloaded, all ISO images show up on the "Custom Image" page of the iRMC S4 web GUI, where admins can activate them either manually or using a timer preset. The image is then mounted as a virtual USB CD/DVD drive, and the system automatically reboots to start from this particular (instead of the standard) device. On PRIMERGY servers that have UEFI Secure Boot and the required keys enabled, it's even possible to boot directly from the SD card.

With a larger selection of ISO images to choose from, eLCM image management gives administrators greater flexibility with regard to both server management and general setup.

PrimeCollect
This functionality was originally developed to provide admins with detailed system status and error information in case of a server malfunction. To this end, ServerView Agents would collect the logging data of a PRIMERGY system in a ZIP archive that could then be sent to Fujitsu's support team for detailed analysis. Until now, this process had to be triggered manually whenever a malfunction occurred.

eLCM-style PrimeCollect makes the entire process much more convenient, as it allows for an automated and scheduled creation of the above-mentioned archives. Like the online update function, this requires the ServerView Agentless Service to be running on the managed server. The archives are then stored on the internal SD card. That way, admins can create an 'archive of archives' that helps them retrieve information about previous server issues and repair options. Moreover, PrimeCollect archives may be sent to another server via the iRMC S4 management LAN or Fujitsu's AIS (AutoImmune Systems®) Connect Service.

eLCM PrimeCollect functionality can be configured and started in the PrimeCollect page of the iRMC S4 web interface; all automatically created archives are accessible via the PrimeCollect archives list for further use. For instance, administrators may define a special reference image that will never be overwritten and could serve as a fallback solution in case of an emergency. Likewise, archives may be deleted or stored as regular files.

Image

Fig. 1: Fujitsu's iRMC S4, the cornerstone of eLCM

Embedded Installation Management (eIM)
As may be easily imagined, the main purpose of this tool is to further simplify and accelerate the initial PRIMERGY setup. Hence, eIM provides an option to run ServerView® Installation Manager (SVIM) out of the box, thus freeing admins of the hassle with traditional installation media. All they need to do is launch the iRMC S4 web interface, navigate to the SVIM PE image that's included on the internal SD card ex factory, proceed to install the host OS, and configure built-in RAID controllers where necessary. What's more, administrators may also apply this method in case they want to perform server maintenance and/or updates on a productive system. To this end, they only need to download a fresh SVIM PE image from a download server to the SD card and proceed from there.

The installation process is typically carried out remotely with the help of the so-called eLCM Profile Management: the administrator picks a suitable installation profile (image) from a pre-defined repository and sends it to the iRMC's remote Restful API. This image contains detailed information about the server setup and configuration. As a result, server administrators can not only trigger a largely automated installation process, but may even run it in unattended mode. Alternatively, it's also possible to kick off the eIM process manually in a local setting.

Embedded RAID Management (eRM)
This feature serves as the eLCM equivalent to the well-known ServerView® RAID Management function that is distributed via conventional installation media. Contrary to what one might expect, this does not mean users will receive a trimmed-down version; instead, eRM retains the full functionality of the 'classic' version. In particular, it supports the same number of RAID levels (8 plus JBOD), allows for concatenation and the creation of RAID volumes, and includes numerous advanced functions like online capacity expansion, data migration between RAID types, consistency checks, and copyback. What's more, just like its counterpart eRM permits administration of various RAID controllers and the devices attached to them, regardless of whether they use SCSI, SATA or SAS interfaces. More details are laid out in the ServerView® RAID Manager user guide.

Embedded Diagnosis/Diagnostics Management (eDM)
Just like eRM, this function is basically the eLCM edition of a familiar ServerView® module, in this case the one that lets users perform online diagnostics while a system is up and running. Put differently, eDM enables sysadmins to carry out performance tests on a variety of hardware components. These tests fall into three broad categories:

  • Standard tests emulate moderate system loads and last for about 10 minutes. These tests follow a predefined pattern and help admins determine how well CPUs, memory modules and HDDs fare under these conditions. For CPUs, the load/idle ratio is set to 10%, and the test is carried out on all CPUs inside the server except for the first (CPU0). Standard memory tests eat up 20% of RAM capacity for the fastest subtests (e.g. "Pattern"); here the CPU load is also at 10%. Finally, standard HDD tests help to check sequential read/write speeds for threads with normal priority; the test covers all logical disks except for the system drive and typically requires 50% of the free disk capacity. Because the demands are rather modest, these tests can be executed without affecting other, productive applications.
  • By contrast, stress tests simulate high system loads and run for two hours on average. Test parameters will trigger the occurrence of RAM bottlenecks as well as simultaneous execution of high-priority tasks and high HDD throughput. Stress tests can be performed on the same components as standard tests, plus optical disk drives (ODDs). For CPUs, the load/idle ratio is set to 50%, and the test runs on all available CPUs (incl. CPU0) at normal priority. Memory stress tests are identical in terms of CPU usage and thread priority; they involve the full set of seven subtests and require 80% of RAM capacity. Given these conditions, both variants will only work if no other applications are running. HDD and ODD stress tests are less demanding, which means applications may remain active: the HDD stress test includes a so-called butterfly test (synthetic random reads), but is otherwise basically the same as the standard test. ODD stress tests consist of two subsets with thread priority set to normal; they will last for 15 minutes and may only be performed if a CD/DVD with at least 500 MB capacity is inserted into the drive.
  • With custom or user-defined tests, administrators are free to adjust test parameters in accordance with individual PRIMERGY servers. They may thus modify specific test characteristics (e.g. CPU load, memory utilization) for more accurate results that correspond with existing usage scenarios.

For more details, please see the ServerView® Online Diagnostics Manual (PDF).

Availability, Licensing and Costs
eLCM is offered as an optional extension for most PRIMERGY models equipped with iRMC S4. To be eligible for eLCM, these systems must feature an SD card socket and "populated" (loaded) SD card. Licenses are available in two dosage forms and are always purchased together with the SD cards: the eLCM Activation License can be ordered with any new, eLCM-ready PRIMERGY model and will be activated ex factory, whereas the eLCM Activation Pack may be ordered for supported systems that are already up and running. Prices vary depending on region.

For detailed information, please refer to the ServerView® specifications and documentation, the iRMC S4 white paper, and the eLCM manual.

Jochen Riedisser

 

About the Author:

Jochen Riedisser

Product Manager, Fujitsu

SHARE

Comments on this article

No comments yet.

Please Login to leave a comment.

X

Please login

Please log in with your Fujitsu Partner Account.

Login


» Forgot password

Register now

If you do not have a Fujitsu Partner Account, please register for a new account.

» Register now