American Megatrends MegaRAC G3 User manual

A Guide to Preparing a Successful
Server Management Solution
MegaRAC G3 allows IT departments to efficiently
implement a policy-based provisioning system
management solution
In today’s economy companies are struggling to keep their revenue
growing at a constant rate; they demand steady productivity from their
employees and 24/7 accessibility from their computer systems. A great
challenge to a company’s IT department lies in keeping its employees
and systems up and running continuously. Read Steve’s IT problem in the
sidebar.
Companies must be prepared. They must have an effective and efficient
provisioning and policy-based system management program in place.
Their program must include more than just a remote management
controller. As Steve realized, he needs more than just a quick fix to a
restart a server after a crash – he needs a complete solution that will
prevent this from happening again, and a solution that will prepare his
company for long-lasting productivity and success.
Why a “quick-fix” doesn’t solve the problem
Assume Steve is going on a Safari trip. He knows there will be
mosquitoes that will give the victim of a bite a deadly disease. To
prepare for the bites he may get, Steve brings medicine that will cure
any disease he may contract from a mosquito bite. However, he is
concerned that he has no medicine that will prevent him from getting
bit in the first place and having to endure the pain and suffering
associated with it.
Just as the fix to curing the mosquito bite doesn’t solve the problem of
the bite, IT administrators face the same dilemma with a traditional
remote management controller or KVM over IP solution. While the
technology may provide a quick fix to restart and reduce the downtime
of a server (low MTTR or Mean Time to Repair), it does not avoid lost
productivity.
To understand how to develop an efficient system management solution,
one must understand all of the terms and features involved in the plan.
System availability
IT administrators must know what the availability of their systems are in
order to determine a total productivity analysis.
PRELIMINARY
STEVE’S SERVER
PROBLEM
A
cell phone rang,
breaking the mood of
Steve’s party. Steve,
who manages the entire
IT department for a big
company, sighed and
answered his phone.
Once again, it was an
alert notification from
his remote management
controller and he knew
that one of his servers
had crashed.
Fortunately for Steve,
he could restart the
server from home –
thanks to his remote
management controller
– and his party could
continue.
Though he was glad he
didn’t have to go into
the office, Steve was
worried about why his
servers had been going
down more frequently
than ever. He wondered
what would happen to
his company’s future
and overall
p
roductivity if this
continued to occur.

Availability = MTTF or mean time to failure
MTTR or mean time to repair
Traditional remote management solutions address the MTTR, and when this is dramatically reduced,
the availability is improved.
Reliability and availability
Though traditional remote management controllers contribute significantly to improving a server’s
availability, they do not play a role in improving the overall reliability of the system. Reliability is a
dependable attribute and is a measure of the system’s continuous service without a failure. If
administrators analyze their server’s failure and predict when it will fail well before it actually does
fail, this will improve the system’s reliability.
Failure management
When reliability and availability reach their highest levels, they automatically achieve the “High
Availability” or HA. A thorough fault management scheme helps to achieve the HA. The fault
management’s functions include:
1. Detection
2. Diagnosis
3. Isolation
4. Recovery
5. Repair
A system and its service
A system is a collection of interacting components that can be a system, subsystem or singular
component. Components are results of the system’s decomposition. A component that cannot be
decomposed is an atomic component. The system’s service is the output of the system to its consumer,
which can be a human or another system. There can be more than one service from one system.
Failures and faults
Failures are the disruption or absence of the system’s service. Faults are the designated root causes of
the failure. A fault can be either active when it generates the error or latent when it has not been
detected.
Hardware v. software failures
When looking at the failures caused by the hardware and software, administrators can see a basic
difference between the nature of the faults in these two domains. Hardware faults are more frequent
with time due to the decay of the system, which typically occurs due the aging of the hardware.
Software faults are caused mainly by the interdependence and logical deficiencies of these modules.
Aging of the system
Hardware ages over time, and it is important for administrators to keep track of how old the hardware
is when trying to detect or prevent a system’s failure. Mechanical and power electronic components
are most affected by aging. Hard drives, power supplies, fans or blowers are more susceptible to
failures than any other component in the system.
Understanding failures
Before an administrator sees the need for provisioning a system management solution, they must
understand the background of their server’s failures and faults. The probability of a failure depends on

the amount of hardware and software in a system. Administrators must take these components into
account when computing the reliability and availability of their managed system.
By interpolating the field data and failure equations, administrators can see the unavailability curve
running from the most pessimistic level to the expected level in overtime. This analysis is facilitated
with the MegaRAC G3 advanced management controller.
Failure prediction
A successful failure prediction improves possible system downtime. Merely monitoring the components
is not enough preparation for a system failure. To properly and accurately predict a failure, an
administrator must conduct a policy-based provision for acquiring data on the monitoring components.
Policy-based provisioning
Policy-based provisioning is the collection of rules an IT department can impose to prevent system
failures and recover the system quickly. These rules must be programmable and applicable to all
possible environments, as different organizations demand different policies.
Policy-based provisioning can be further explained with an example: An imaging company stores and
uses massive storage data every day with its storage data servers. At the beginning of the day, the
administrator first monitors the server for data reliability and keeps track of the drive cage
temperature and airflow. Due to the exhaustive usage of the hard drives, the cage’s temperature rises
faster than that of any other server. While the temperature rises, the airflow system, consisting of fans
and blowers, must be turned off or slowed down to lengthen its life. To prevent this from happening
daily, a proper policy must be reinstated that will tell the server to turn on the drive cage fans when
the temperature begins to rise. This will prevent faults, errors and eventually failures.
A provisioning policy can replace the situation in the example above. The policy will allow the
administrator to set a condition to turn the drive cage fan on, based on the historical rising patterns of
the specific drive cage temperature. These policies must be intelligent in nature and programmable to
adapt to the environment.
Power policies
An important policy is a power policy, which includes a regularly scheduled power cycle of the
computers that can save energy and refreshing the memory usage and planned maintenance of the
system. Software components, depending on the nature in which they are used, may consume more
memory. Over a period of time, the memory’s utilization begins to increase and the system becomes
less responsive. This scenario is typical in the IT industry and a quick power cycle can recover the
desired productivity. These policies can be easily deployed with the MegaRAC G3.
MegaRAC G3 and the traditional RAC
A traditional remote access controller (RAC) or service processor normally monitors the hardware
components and a static comparison of the predefined thresholds. The MegaRAC G3 incorporates a
unique methodology to extend the manageability of the system through its Platform Development Kit
(PDK). The G3’s PDK provides OEMs and integrators with the ability to build a policy-based provisioning
software component and user interface. With this, an OEM can add its own intellectual properties
based on the management features or special hardware features.
AMI is committed to extend the provisioning tools built into the MegaRAC G3 in future releases, which
will give end users the ability to add simple rules to manage their servers.

MegaRAC G2 v. MegaRAC G3
MegaRAC G3 provides a management infrastructure or platform that can be extended limitlessly into
the future. The MegaRAC G2 and older generations offer a non-extendable pre-specified solution. The
MegaRAC G3’s main strength is extendability through the G3/PDK that allows many modules and
utilities to be developed and take care of the management policies that could not be done in the past.
The main differences between the two cards are described below:
1. Provisioning: this is the strongest feature that can be developed based on the G3 infrastructure.
2. Processing power: MegaRAC G3 uses a much higher-powered processor (400 MIPS), compared to 40
MIPS in the MegaRAC G2.
3. Memory: the processing power of the G3’s architecture allows more memory (up to 2 GB) in the
future (it currently has 32 MB installed). More features, functionalities and utilities can be
developed for installation for future releases.
4. Reliability and availability (RAS): with provisioning software (developed separately on the G3),
both the system’s reliability and availability can be addressed.
5. Tools – G3/PDK: the PDK provides the ability to develop and deploy software and utilities that can
run on the MegaRAC G3.
6. USB CD and floppy: a simultaneous presence of two virtual devices (CD and floppy) is unique for
the administrators.
7. Image-based boot: MegaRAC G3 can boot from a bootable image located at the client system.
Administrators can centralize the repository of all the bootable images needed for all servers.
8. IPMI style: MegaRAC G3 follows an IPMI-style event log management and alert mechanism that
comply with industry standard features.
9. RMCP: this allows for a common method to interact with the BMC in the system over a secured
TCP/IP connection. This helps tremendously in blade server management.
10. PEF and PET: Platform Event Filtering (PEF) and Platform Event Trap (PET) are two standards
defined by IPMI and are followed in the MegaRAC G3. This makes policy-based provisioning
software development easier for alert notifications and event logs.
11. SSH-based command line interface: this provides a command line interface (CLI) to the MegaRAC
G3, using a secured shell. Unix/Linux administrators can easily use this CLI from other Unix
systems.
12. Firewall: for the highest security purposes, MegaRAC G3 has a built-in firewall capability that
allows administrators to block desired TCP ports
13. User interface: the GUI has been improved and is based on the G3/PDK, which can be extended or
easily customized.
14. Multi-language support: MegaRAC G3 can support multiple languages, as it supports Unicode
(UTF8).
Conclusion
To understand the real capabilities and strengths of the MegaRAC G3, one final example is given.
Suppose someone purchased an electronic organizer that kept track of telephone numbers, addresses,
appointments. While these features were what they were looking for, they were limited. The person
did not purchase a Personal Digital Assistant (PDA), which did not have many features, but did have the
potential to offer more than the electronic organizer in the future in regards to the numerous utilities
and software that will be able to be deployed and run on the PDA.
This situation can be used to describe the MegaRAC G3. While it provides remote access features like
KVM/IP, remote power controls, remote monitoring, etc., it can also deploy and run future utilities
developed by its strong G3/PDK. In addition, several important management utilities will be written in
the future by AMI that will be specially designed for the MegaRAC G3.