Exam Reviews
Clustering Services
Hardware know-how gets its day in the sun with the Clustering exam. Word of warning: This test isn’t for the faint of heart.
Well, it’s about time! Microsoft has finally noticed
that hardware does play an important role in an enterprise.
This test is living proof of that. Not only do you have
to know Windows 2000 Advanced Server to pass, you also
need more than a passing familiarity with the different
types of hardware supported by Microsoft Clustering Services.
Clustering
Services, 70-223 |
Reviewer’s Rating:
“This exam should be subtitled, ‘Everything
you wanted to know about SCSI, Fibre
Channel, and other assorted hardware—but
didn’t want to ask.’ Not for the beginner.
Know your clustering solutions cold.”
Title: Installing,
Configuring, and Administering Clustering
Services by Using Microsoft Windows
2000 Advanced Server
Current Status:
Went live in November 2000.
Who Should Take
It? MCSE elective. Candidates
should have a minimum of one year’s
experience installing, configuring,
and administering a network operating
system that includes SCSI and Fibre
Channel devices.
What Classes
Prepare You? None. This one has
to be on experience alone.
|
|
|
True to the format of the Win2K exams, this test has
verbose questions and expects you to know the minute details
of the product. If you don’t read carefully, you’re liable
to select an answer that’s subtly different from the correct
one. The test also gives detailed, real-life questions
that are enterprise-based. Remember that the Win2K exams
are designed for people who have a year or more of experience,
so this challenge isn’t for the beginner.
I also recommend being a Compaq ASE or having equivalent
training before tackling this one. If you find yourself
weak in the hardware arena, this test isn’t for you. You’ll
find yourself at a distinct disadvantage if you don’t
understand how SCSI, Fibre Channel, and hardware RAID
arrays work. Make sure you have a strong background in
all of these topics before testing.
Preinstallation Pitfalls
What’s clustering, you ask? Clustering is the process
of installing applications on a shared drive or array
between two or more machines, called nodes. These nodes
are backups for one another in case the other machine
fails. If one machine fails, the other immediately takes
over for it. This process is called failover. When the
original machine is back online and control is returned
to it, a process called failback takes place.
Before you can even get out of the gate and install the
Clustering Service, you need to know how to configure
Win2K Advanced Server. The first thing to understand is
what network settings are appropriate. Cluster servers
need to be multihomed. In other words, they typically
have more than one network card installed. The first network
card is used for communication on the public network and
the second for communication between nodes on a cluster.
You should know how to configure Win2K Advanced Server
to support the multihomed configuration. Be familiar with
what protocols need to be installed on which NICs under
specific circumstances. Know how to provide fault tolerance
when a network card, switch, router, or hub fails. Be
able to assess an existing network for utilization problems
and know how to improve network communication performance
between nodes on a cluster.
Even though this is a test on Clustering Services, knowing
the “mundane” network services like DNS and WINS will
go a long way. Sometimes problems are as simple as knowing
that your particular variety of name resolution isn’t
appropriately working. On occasion, a client might not
have the correct IP address or subnet mask to communicate
with a cluster. Or perhaps a client couldn’t get a lease
from a DHCP server, and it configured itself using Automatic
Private IP Addressing. All these things are possible problems
that could easily happen to clients on a network. True
enough, they don’t have a whole lot to do with clustering,
but they can certainly affect overall connectivity.
Before you can install the Clustering Service, you need
to have physical connectivity to your quorum resource,
which—in layperson terms—is your external shared SCSI
drive or array. Be able to troubleshoot issues that involve
connectivity to Fibre Channel hardware. Know where in
your array terminators are placed and whether or not you
should enable passive termination of the SCSI chain. Also
know what SCSI IDs the quorum resource should have and
how the devices should be configured if you have both
internal and external SCSI devices.
Tip: I’d recommend getting
a good hardware book and learning all about SCSI and Fibre
Channel. If you go into the test without understanding
the hardware aspect of the exam, you’re going to have
a difficult time.
Configuring Clusters
Of course, installation and configuration are the meat
and potatoes of any clustering solution. You need to be
able to install the Clustering Service on both nodes of
a Win2K Advanced Server cluster. It’s important to know
how to set up a cluster to work in a domain environment.
For example, both nodes of a cluster need to be in the
same domain in order to work. It can’t hurt to know what
a service account is and what specific rights that service
account needs.
Some more basic Win2K understanding on disk configuration
can’t hurt either. If you’re already Win2K-certified,
I’m sure you know the difference between basic and dynamic
disks. Remember, a “stripe set with parity” is what we
called software RAID on Windows NT 4.0, but a “RAID 5
volume” is what we call the same thing on a Win2K installation.
Consider the fact that nodes have internal hard drives
or arrays in addition to the quorum resource. Figuring
out which file system and fault-tolerance scheme should
be present on internal arrays and external arrays is also
a good skill.
In addition to knowing how to install a fresh copy of
Win2K Advanced Server and the Clustering Service, be familiar
with the upgrade process from NT 4.0 Enterprise Edition
to Win2K Advanced Server. Know what issues are associated
with the upgrade process and how to resolve anomalies
that might occur due to the upgrade.
Tip: Brush up on your Win2K
disk-management skills.
Application Awareness
Once the cluster server is installed and configured appropriately,
you can take the action of installing your applications.
I’m sure you’re thinking, “Oh, this is easy, I can install
applications.” Well, it really isn’t as easy as you might
think.
First, know if your application is cluster-aware or not.
A cluster-aware application is specifically programmed
with custom functionality to support clustered installs.
After that, you run the application’s installation, but
you have to know whether the application gets installed
on the quorum device or on the local node’s hard drive.
You also have to know where the application’s data files
go. It can be a very detailed process.
When installing a cluster-aware application, you need
to know what resources the application is dependent upon.
Some common examples of resources would be an IP address,
a network name, a physical disk, and so forth. Once the
cluster server is told what resources an application needs,
it’s able to know what to monitor for failure. For example,
if a network card fails, the IP address is inaccessible;
failover to the secondary node can then take place. You
should know what resources are needed and how to add them
to an application’s configuration. You should also understand
what resources are dependent upon other resources.
It’s a good idea to be familiar with the installation
of several Microsoft cluster-aware applications. In Win2K,
it’s now possible to cluster several common networking
services, like DHCP and WINS. You should know how to set
up, configure, and troubleshoot issues that involve these
two services. You should also be able to manage a cluster
that has IIS installed.
In addition, Distributed File System (Dfs) can be clustered
under Win2K—know which type of Dfs (domain-based or standalone)
is appropriate under the circumstances. Also, Win2K has
the ability to cluster printing services. Be familiar
with how to install and configure these items on your
cluster server.
Once everything is installed and configured, it’s a perfect
world, right? Of course not. The core of our jobs as network
administrators and engineers is to be able to troubleshoot
any problems that might arise. Know how to start the cluster
service from the command line along with all of the associated
options. This is usually used to start the services under
special conditions when the clustering service won’t start
due to a problem.
Tip: Be sure you know where
the cluster logs are located and how to manage them.
Management Mayhem
Okay. So now our applications are installed and running
like a well-oiled machine. Now what? Well, we have to
be able to handle day-to-day management, don’t we? Daily
management includes delegating access; handling backups;
and being able to handle hardware maintenance, like adding
new disks to a cluster.
Be sure you know how to replace a node in case of a total
hardware failure. Machines don’t always fail totally,
however, so you should be able to replace a failed SCSI
disk or RAID array, if needed. Know how to troubleshoot
problems that might arise due to a failed quorum resource.
Also be familiar with how to remove a node from a cluster
and add another.
When a failover occurs, it should be your intent to repair
the primary node and return control back to it. If you’re
not careful, you can end up attempting to failback before
the machine is ready. Know how to configure failover and
failback options to avoid this.
Avoiding Accidents
Many things can go wrong during an installation of the
Clustering Service. Be able to recover from internal and
external disk failures and know how to replace a failed
SCSI drive or array. Believe me, if a failure happens,
your boss is going to jump on you to fix the problem.
Avoiding disaster is really important, too. Know how
to schedule and perform backups of both the data on your
quorum resource and your internal drives and arrays. Understand
which parts of the Win2K operating system need to be backed
up for a successful restoration of a cluster and know
how to perform reconfiguration if your backup isn’t complete.
In addition to knowing how to back up and recover, be
sure you know how to make the existing system run better.
Be able to optimize internal and cluster network traffic
to best suit the needs of the organization. Also, understand
how to distribute load between clusters, if appropriate,
and be able to pick which node in the cluster should be
the owner of the application.
Experience is the Key
Like I mentioned, this is certainly not an exam for the
faint of heart. I found few study materials for it and
no Microsoft Official Curriculum for Win2K clustering.
Your best bet (as always) is to have experience with the
technologies involved and go from there. Make sure you
spend the time to read each question carefully to get
all the details. One thing this exam definitely requires
is attention to detail. Good luck!