My Top 10 Tips for Passing the Server+ Exam
Server+ exam preparation author and regular CertCities.com contributor Robert Bogue shares his insight on the what you need to know to pass this CompTIA exam.
by Robert L Bogue
1/21/2002 -- Since its launch last year, CompTIA's Server+ has become a prerequisite or elective for many vendor-specific certification programs, including Compaq, HP and Novell. But it took Microsoft's recent announcement that Server+ -- when combined with A+ -- would count as an elective for the company's new Microsoft Certified Systems Administrator (MCSA) title for the buzz to officially swell.
Whether you're looking at Server+ as a stepping stone into another program or as a standalone achievement, there's a lot you'll need to cover to pass this exam. In a lot of ways, Server+ title is the big brother of A+: Everything that A+ challenges you on, for the most part, you'll get on the Server+, since it is so hardware-centric. At least in that respect, this exam is somewhat predictable. However, there are some tricks on Server+ that A+ doesn't cover. Be sure to get the test objectives from CompTIA's Web site so you know everything you'll be asked about. From there, I offer you these tips for getting past some of the exam's obstacles.
1. Document, Document, Document
One of the biggest differences between the A+ exam and the Server+ exam is the expectation that you're not the only one working on a problem. In a desktop PC world you might be the only person resolving problems or making changes to a PC. However, in a server environment this is probably not a good assumption. Server+ approaches many questions from the point of view of a large organization that has multiple people maintaining a server farm (multiple servers). As such, instead of taking action straight-away on a particular problem, your first step might be to review server logs, looking for recent network changes.
2. Know Your Backups
CompTIA expects you to know about tape backups in particular and disaster recovery in general. Let's review three main types of backups that you can perform:
- Full -- All files are backed up, and the archive bit is cleared.
- Incremental -- All of the changed files are backed up, and the archive bit is cleared.
- Differential -- All of the changed files are backed up, and the archive bit is not reset.
Differential backups take more time than incremental backups on a daily basis because they will back up the data that was changed since the last full (or incremental) backup. However, if you're using differential, it takes only the last full backup and the last differential backup to completely restore a system. Using incremental, the last full backup and every incremental backup must be used to restore a system.
You may also want to do some reading about disaster recovery services and the differences between Hot, Warm and Cold (if you don't know, look it up!).
3. SCSI Isn't Skuzzy
Yes, our friend SCSI -- you'll find in him various places on the exam objectives. Let's review some basics:
- Each end of a SCSI chain must be terminated, and it should never be terminated in the middle.
- The SCSI adapter is only the end of the SCSI chain when there are devices on either the internal or external interface -- but not both. Never turn termination on the SCSI adapter on when there are devices on both the internal and external bus.
- Each SCSI device, including the adapter, has an ID. Adapters are typically on ID 7, but not always. For "narrow" SCSI you can have IDs 0-7, and for "wide" SCSI you can have IDs 0-15. So "narrow" SCSI can have 7 devices other than the adapter, and "wide" SCSI can have 15 other than the adapter. (Narrow SCSI is any version from SCSI-2 and prior. Wide is SCSI-3 and higher including Ultra2, Ultra160 and Ultra320.)
- There are two kinds of SCSI buses in widespread use today. The most popular is Single Ended (SE) and is often on older devices. The more recent standard (required for Ultra2, Ultra160 and Ultra320) is Low Voltage Differential (LVD). Most LVD devices can "fall back" into a SE mode. No damage will occur by connecting LVD drives on an SE bus, or vice versa.
4. RAID 'Em Dead
Redundant Array of Inexpensive Disks (RAID) is another area covered on this exam's objectives. (By the way, RAID is sometimes expanded to Redundant Array of Independent Disks, but the original proposal read "Inexpensive.") Anyway, let's quickly take a look at the three levels of RAID:
- RAID 0 -- Commonly called striping. RAID 0 allows two or more physical disks to be combined into one logical disk. This version of RAID is used for performance reasons. It performs better than a single physical drive for both reads and writes. RAID 0 does not provide fault tolerance, so if you lose one drive, you lose all information on all drives.
- RAID 1 -- Commonly called mirroring (or duplexing). RAID 1 takes two (and only two) physical disks to form one logical disk of the same size as one of the physical disks, providing fault tolerance. Information stored on the logical disk is stored identically on both disks. Read performance is better but write performance is slightly less than a standard drive. This configuration has the drawback of cutting your storage space in half.
- RAID 5 -- RAID 5 uses a rotating parity disk to provide redundancy while minimizing the amount of physical disk space required. You must have at least three physical disks. The logical disk created is the size of (x-1)*n where x is the number of physical drives and n is the size of the disks. For example, if you have 6 four GB physical hard drives in a RAID 5 array it will yield 20GB of storage space (6-1 = 5, 5 x 4GB = 20GB). RAID 5 parity calculations are generally done in hardware but can be done by operating systems such as Windows NT and Windows 2000 with an associated CPU requirement. RAID 5 generally performs slightly better for writes and better for read operations. The parity information that is stored in the array provides the fault tolerance for a single drive failure.
Also be aware of RAID 0+1 and RAID 1+5. These are combinations of the three RAID levels above. For instance, RAID 0+1 allows the use of multiple drives in a mirrored configuration where the information is always written to two different drives.
5. Is Power a Problem?
As a Server+ professional, CompTIA expects you to be able to troubleshoot power-related problems. When thinking through problem scenarios, make sure that you sconsider how each device is getting it's power. You may find that the problem is as simple as a server or device not working because it doesn't have any power.
6. Name That Bottleneck
Performance monitoring is a major part of any server administrator's job. You should know that memory problems sometimes show up as disk problems because of excessive virtual memory paging operations. Always look for memory as a culprit first. Excessive paging can also be the cause of the performance problem. Next look for processor problems, such as sustained processor activity above 80 percent. Use long polling intervals for measuring processor time so that you don't accidentally interpret a peak processing period as a problem. Finally, look at disk activity: Sustained utilization at 100 percent or physical disk queues that exceed two entries are cause for concern.
7. Beware the IDEs of March
Although CompTIA puts more focus on SCSI, do know the basics of IDE drives. For example, did you know that ATA/66 and faster drives require a hybrid IDE cable that has the standard 40 pin connector but uses 80 conductors (twice that of a normal cable) to conduct signals from the drive to the motherboard?
8. Rack 'Em Up
Racks aren't a major focus of this exam's objectives, but you should be aware of the basics, like the following:
- Racks are measured in units called 'U' that are 1.75 inches tall.
- Most racks are a standard 19 inches wide.
- Racks require special power and cooling considerations because of the high density of servers. Many racks may require more than one 15 AMP power circuit. They may also require booster fans to force air up and out of the rack.
9. OS/2 (No, Really?)
Yes, OS/2 is listed in the exam blueprint. I can't tell you to ignore it -- my editor won't let me -- but I can tell you what I did, which is ignore it completely, and my score didn't suffer one bit.
Interestingly, the job task analysis (JTA) on which the exam is based clearly revealed that no one cares about OS/2 in their environment. However, the objective was left in (maybe because IBM is a cornerstone certification committee member for this exam?).
10. Select Your Training Material Carefully
As with many exams, there are a wide variety of training options available -- check each out thoroughly before you decidewhat to buy. For Server+ books, the current selection runs the gambit from well-presented, error-free sources to materials that are badly organized and littered with mistakes. Of course, I'm completely biased toward my own book, but there are others out there, as well as training classes and practice exams. No matter what you choose, ask around for other user's opinions before you plunk down money for anything.
In general, if you've had to work with server hardware, you should have no problems. If you're new to the idea of SCSI, RAID and other server-centric hardware, take your time to become really familiar with these server-based technologies. And again, it doesn't hurt to pass CompTIA's A+ Hardware exam first. Good Luck! 
Robert L. Bogue is an MCSE (NT 4.0 & Win2K), MCSA, etc. He runs Thor Projects, a systems consulting company in Carmel, IN. Robert is also the lead author of Que Certification's MCSA Training Guide (70-218): Managing a Windows 2000 Network (ISBN 0-7897-2766-8, Available June 2002). If you want to learn more about Robert's other books go to http://www.thorprojects.com/author.
More articles by Robert L Bogue:
|
