Server+ Certification Training Kit by Microsoft Corporation, Microsoft

From the Publisher

The Server+ credential is the newest certification developed by CompTIA, the Computing Technology Industry Association, and this self-paced training kit provides in-depth exam preparation for IT professionals moving up from entry-level or A+ status. The Server+ Certification Kit delivers a thorough, vendor-neutral study of hardware-related issues in the networked environment, including RAID, SCSI, multiple CPUs, and hot swapping. Service managers, system engineers/administrators, help desk staff, and other mid-level and upper-level technicians can use the kit to build real-world expertise as they prepare for the corresponding skill areas of the Server+ exam. Topics include installation, configuration, upgrading, proactive maintenance, environment, troubleshooting, and disaster recovery. The kit is modular and self-paced, with hands-on, skill-building exercises. And the entire book is featured on CD-ROM for easy searches and reference.

Positioning Statement:

Hands-on, self-paced training for Server+ Certification the next step up from A+

Unique Selling Proposition:

Applies the reliable, competency-based learning format of Microsoft® training kits to CompTIA Server+ exam preparation

Key Book Benefits:

Delivers hands-on learning for material covered by the CompTIA Server+ exam, which certifies expertise with hardware in the networked environment Supports exam preparation and skills transfer to the job in a single kit Features the entire book on CD-ROM for easy searches and reference Will Sell Like.

Product Details

• Table of Contents
• Read an Excerpt
• Read a Sample Chapter

Table of Contents

Read an Excerpt

Chapter 13: Upgrading UPSs, Monitoring the System, and Choosing Service Tools

About This Chapter

The lessons in this chapter will prepare you for the next few chapters, which focus on proactive maintenance of the system.

Before You Begin

• Your server documentation
• Any batteries or UPSs necessary to replace existing batteries and UPSs

Lesson 1: Upgrading UPSs

• Add or replace batteries on a UPS
• Install new inverter units on matrixed UPSs

Replacing UPS Batteries

This discussion focuses first on the mid-size UPSs that have replaceable batteries. Unless you have a large environment, you're likely to have this type of UPS, with batteries housed inside the unit.

Replacing a Battery in a Mid-Sized UPS

The primary challenge when replacing a battery in a mid-sized UPS is that you have to open the unit itself up to do so. In most cases, the UPS manufacturer will indicate that the unit is designed to allow hot swapping of the battery. That is, the UPS (and the servers it supports) need not be shut down to replace the battery.

However, the manufacturer will also generally include some strong cautions about doing a hot-swap upgrade because of the high voltages present inside the UPS. These cautions should be sufficient to convince you that it's best to plan to replace the UPS battery during scheduled maintenance. The risk of personal injury and the possibility that the UPS might shut down during this process are substantial enough to warrant replacing the battery while the connected servers are shut down.

Mid-sized UPSs use sealed lead-acid batteries similar to the ones used in cars. Unlike some car batteries, however, UPS batteries are completely maintenance free. The fact that these batteries are sealed helps ensure that no harmful acid leaks out inside the battery case.

A general process for replacing the batteries in a UPS appears below. This procedure is for reference only. Due to the potentially dangerous voltages present inside a UPS, please refer to the battery replacement instructions for your specific equipment.

To replace the batteries in a UPS

1. Shut down all servers and equipment connected to the UPS.
2. Shut down the UPS.
3. Unplug the UPS from the wall outlet.
4. Remove the new battery from its packaging. (Save the packaging for returning the old battery.)
5. Remove the front cover or case as indicated in the battery replacement guidelines.

---

CAUTION:
Even though power is disconnected from the wall, the batteries and circuitry in the UPS still contain significant electrical energy. You may want to remove metal rings and bracelets and be cautious of contact with bare wires.

---

6. Remove the screws that hold the battery compartment door in place.
7. Slide the battery forward until both wires connecting the battery to the UPS are visible.
8. Disconnect the red, positive wire from the battery.
9. Disconnect the black, negative wire from the battery.
10. Slide the battery the rest of the way out.
11. Slide the new battery in a little, leaving both terminals exposed.
12. Connect the black, negative wire to the battery.
13. Connect the red, positive wire to the battery.
14. Ensure that no metal is exposed on the wire or the battery terminals.
15. Push the battery completely into the battery compartment.
16. Close and secure the battery door with the screws you removed earlier.
17. Replace the UPS cover.
18. Plug the UPS back into the wall outlet.
19. Turn on the UPS, and ensure that the bad battery indicator is not lit. If the bad battery indicator is lit, call the UPS manufacturer or follow the troubleshooting procedures in the UPS user manual.
20. Pack the old battery in the packaging you saved from the new battery, and seal it for return to the vendor's battery recycling program.

As you can see, the process of installing a replacement battery in an existing unit is relatively trivial. The only significant issue involves the need to exercise caution because of the potentially high voltages remaining in the UPS.

Replacing a Battery in a Large UPS

Once UPSs reach a certain size, the inverter—the part that converts the battery power to alternating current (AC)—is separated from the batteries. This separation allows you to replace the batteries without opening the inverter cabinet, helping to reduce the risk of shock when replacing batteries.

The size at which the inverter and batteries are separated varies by manufacturer and even by product line within a manufacturer. However, most of the time the change from internally sealed batteries to external batteries happens at around the 3 KVA threshold. UPSs that supply more than 3 KVA of power tend to have their inverter kept separate from the batteries, in part to keep the heat that the inverter generates away from the batteries, which don't respond well to heat.

For most larger UPSs, the procedure for replacing external batteries is simple. The following steps describe a typical UPS battery replacement when the UPS uses external battery packs supplied by the manufacturer. This type of arrangement is generally used in so-called matrixed UPSs (discussed in a later section) and in UPSs that are just large enough to warrant external battery cabinets but are too small to require a dedicated rack of batteries.

To replace a UPS battery when the UPS uses external battery packs supplied by the manufacturer

1. Shut down the connected servers and devices (optional).
2. Shut down the UPS (optional).
3. Disconnect the battery cable from the UPS. You may have to remove a locking mechanism.
4. If you're removing more than one battery pack, disconnect each battery pack from the others.
5. Remove the battery pack(s).
6. Place the new battery pack(s) next to the UPS.
7. Connect the new battery pack(s) together, and lock the connectors in place (if a locking mechanism is provided).
8. Connect the first battery pack to the UPS, and lock the connector in place.
9. Verify that the battery failure indicator light on the UPS is off.

Larger UPSs require complete racks of batteries. Service technicians most often install these batteries to ensure that the wiring is done properly and that proper safety precautions are followed. If for some reason you decide to replace the batteries yourself, you'll follow a procedure similar to the following. As always, read the UPS documentation and follow the replacement instructions for the specific UPS.

To replace batteries in a large UPS

1. Lock the UPS into line-conditioner-only mode, if possible. This will allow the UPS to continue using its transformers and MOVs to suppress power surges, along with preventing it from trying to transfer to battery power while the replacement is in progress.
2. Disconnect the batteries from the UPS by removing the main connector from the UPS to the batteries.
3. Disconnect each battery connecting cable, starting at the UPS and working backward. Place a protective plastic cap over the battery terminal after the cable is removed. Note the connection order, and be careful not to make contact between the positive lead of the battery and either the rack or a negative lead.
4. Remove each used battery from the rack.
5. Install each new battery into the rack.
6. Individually connect each battery, starting with the battery farthest from the UPS and moving forward. Be careful not to touch any positive terminal or cable with any negative terminal, cable, or the rack itself.
7. Once you've connected all of the individual cables, reconnect the battery racks to the UPS.
8. Verify that the UPS sees valid batteries attached and that it doesn't indicate any kind of an error.

Adding New UPS Batteries

The second difference involves the need to make sure that the battery cables can handle the current they will carry. When you're simply replacing batteries, you know that the connecting cables are sufficient to carry the current from the batteries to the inverter. When you're adding batteries to an existing configuration, however, you need to verify that the battery cables connecting the batteries to the inverter are sufficient to carry the current.

There are guidelines for the size of cable required for a given voltage and amperage, but the best idea is to contact the UPS manufacturer and verify that the cables are sufficiently sized for the additional batteries.

Working with Matrixed UPSs

The fundamental difference between a matrixed UPS and a mid-sized UPS with replaceable batteries is that in a matrixed UPS the power generation and distribution components can be replaced independently. This means that it automatically supports a built-in fail-safe bypass mode of operation when the inverter is nonfunctional. In other words, the power will continue to be conditioned by the transformers and MOVs even if the inverter is not present.

This ability is certainly an advantage, but the sales of matrixed UPSs have not been significant, so similarly sized UPSs with replaceable batteries are usually a better deal than matrixed UPSs. (Due to economies of scale, devices that are sold in quantity are generally cheaper than those that cannot be made in similar quantities. This leads to matrixed UPSs being more expensive.)

Replacing an Entire UPS

Replacing a large, hardwired UPS with a new one is a simple process. (Most large UPSs are hardwired into the electrical system of the building rather than being plugged in.) The new wiring is run for the new UPS, the supported systems are shut down, the old unit is shut down, the new unit is connected, and the new unit is started. The biggest catch to this whole procedure is the new wiring. Because the new UPS will support greater loads, it will require more power to run through it. As a result, the wires connecting the UPS to the outside utility and to the internal distribution panel often need to be upgraded. In addition, the internal distribution panel itself may need an upgrade. Take these electrical changes into account when you plan for the new UPS.

Lesson Summary

Read a Sample Chapter

• About This Chapter
• Before You Begin
• Lesson 1: Upgrading UPSs
• Replacing UPS Batteries
• Adding New UPS Batteries
• Working with Matrixed UPSs
• Replacing an Entire UPS
• Lesson Summary
• Lesson 2: Monitoring the System
• System Monitoring Hardware
• System Monitoring Software
• Practice: Select an Appropriate Monitoring Solution
• Lesson Summary
• Lesson 3: Selecting Service Tools
• Essential Hardware Tools
• Software Tools
• Lesson Summary
• Review

About This Chapter

The past several chapters have talked about upgrading the hardware within a server. This chapter finishes our discussion of upgrades by talking about upgrading the external equipment that most professionals forget about: the UPSs. It also describes the various software and hardware approaches to system monitoring. Finally, it lists a number of service tools that may help you maintain your existing systems and identify when an upgrade is necessary.

The lessons in this chapter will prepare you for the next few chapters, which focus on proactive maintenance of the system.

Before You Begin

To complete this chapter, you should have

• Your server documentation
• Any batteries or UPSs necessary to replace existing batteries and UPSs

Lesson 1: Upgrading UPSs

This lesson discusses the various ways of upgrading a UPS, including replacing existing batteries and adding new batteries to increase run time. It also discusses matrixed UPSs and how to upgrade the inverter units in addition to the batteries.

• Add or replace batteries on a UPS
• Install new inverter units on matrixed UPSs

Estimated lesson time: 20 minutes

Replacing UPS Batteries

UPS batteries have a fixed useful life, after which they are unable to provide the power the UPS inverter needs to support the attached load. Thus, all batteries eventually need to be replaced. Battery life is generally listed with the UPS but can range from 18 months to 5 years, depending upon the type, size, and usage of the battery. As you learned in Chapter 2, Lesson 2, there are three types of UPSs. The small UPSs have fixed batteries that aren’t replaceable. In these UPSs, you must replace the UPS itself at the end of the batteries’ life span. Mid-sized UPS have replaceable internal batteries, and large UPSs have external battery racks that house replaceable batteries.

This discussion focuses first on the mid-size UPSs that have replaceable batteries. Unless you have a large environment, you’re likely to have this type of UPS, with batteries housed inside the unit.

Replacing a Battery in a Mid-Sized UPS

The primary challenge when replacing a battery in a mid-sized UPS is that you have to open the unit itself up to do so. In most cases, the UPS manufacturer will indicate that the unit is designed to allow hot swapping of the battery. That is, the UPS (and the servers it supports) need not be shut down to replace the battery.

However, the manufacturer will also generally include some strong cautions about doing a hot-swap upgrade because of the high voltages present inside the UPS. These cautions should be sufficient to convince you that it’s best to plan to replace the UPS battery during scheduled maintenance. The risk of personal injury and the possibility that the UPS might shut down during this process are substantial enough to warrant replacing the battery while the connected servers are shut down.

Mid-sized UPSs use sealed lead-acid batteries similar to the ones used in cars. Unlike some car batteries, however, UPS batteries are completely maintenance free. The fact that these batteries are sealed helps ensure that no harmful acid leaks out inside the battery case.

A general process for replacing the batteries in a UPS appears below. This procedure is for reference only. Due to the potentially dangerous voltages present inside a UPS, please refer to the battery replacement instructions for your specific equipment.

To replace the batteries in a UPS

1. Shut down all servers and equipment connected to the UPS.
2. Shut down the UPS.
3. Unplug the UPS from the wall outlet.
4. Remove the new battery from its packaging. (Save the packaging for returning the old battery.)
5. Remove the front cover or case as indicated in the battery replacement guidelines.

---

CAUTION:
Even though power is disconnected from the wall, the batteries and circuitry in the UPS still contain significant electrical energy. You may want to remove metal rings and bracelets and be cautious of contact with bare wires.

---

6. Remove the screws that hold the battery compartment door in place.
7. Slide the battery forward until both wires connecting the battery to the UPS are visible.
8. Disconnect the red, positive wire from the battery.
9. Disconnect the black, negative wire from the battery.
10. Slide the battery the rest of the way out.
11. Slide the new battery in a little, leaving both terminals exposed.
12. Connect the black, negative wire to the battery.
13. Connect the red, positive wire to the battery.
14. Ensure that no metal is exposed on the wire or the battery terminals.
15. Push the battery completely into the battery compartment.
16. Close and secure the battery door with the screws you removed earlier.
17. Replace the UPS cover.
18. Plug the UPS back into the wall outlet.
19. Turn on the UPS, and ensure that the bad battery indicator is not lit. If the bad battery indicator is lit, call the UPS manufacturer or follow the troubleshooting procedures in the UPS user manual.
20. Pack the old battery in the packaging you saved from the new battery, and seal it for return to the vendor’s battery recycling program.

As you can see, the process of installing a replacement battery in an existing unit is relatively trivial. The only significant issue involves the need to exercise caution because of the potentially high voltages remaining in the UPS.

Replacing a Battery in a Large UPS

Once UPSs reach a certain size, the inverter—the part that converts the battery power to alternating current (AC)—is separated from the batteries. This separation allows you to replace the batteries without opening the inverter cabinet, helping to reduce the risk of shock when replacing batteries.

The size at which the inverter and batteries are separated varies by manufacturer and even by product line within a manufacturer. However, most of the time the change from internally sealed batteries to external batteries happens at around the 3 KVA threshold. UPSs that supply more than 3 KVA of power tend to have their inverter kept separate from the batteries, in part to keep the heat that the inverter generates away from the batteries, which don’t respond well to heat.

For most larger UPSs, the procedure for replacing external batteries is simple. The following steps describe a typical UPS battery replacement when the UPS uses external battery packs supplied by the manufacturer. This type of arrangement is generally used in so-called matrixed UPSs (discussed in a later section) and in UPSs that are just large enough to warrant external battery cabinets but are too small to require a dedicated rack of batteries.

To replace a UPS battery when the UPS uses external battery packs supplied by the manufacturer

1. Shut down the connected servers and devices (optional).
2. Shut down the UPS (optional).
3. Disconnect the battery cable from the UPS. You may have to remove a locking mechanism.
4. If you’re removing more than one battery pack, disconnect each battery pack from the others.
5. Remove the battery pack(s).
6. Place the new battery pack(s) next to the UPS.
7. Connect the new battery pack(s) together, and lock the connectors in place (if a locking mechanism is provided).
8. Connect the first battery pack to the UPS, and lock the connector in place.
9. Verify that the battery failure indicator light on the UPS is off.

Larger UPSs require complete racks of batteries. Service technicians most often install these batteries to ensure that the wiring is done properly and that proper safety precautions are followed. If for some reason you decide to replace the batteries yourself, you’ll follow a procedure similar to the following. As always, read the UPS documentation and follow the replacement instructions for the specific UPS.

To replace batteries in a large UPS

1. Lock the UPS into line-conditioner-only mode, if possible. This will allow the UPS to continue using its transformers and MOVs to suppress power surges, along with preventing it from trying to transfer to battery power while the replacement is in progress.
2. Disconnect the batteries from the UPS by removing the main connector from the UPS to the batteries.
3. Disconnect each battery connecting cable, starting at the UPS and working backward. Place a protective plastic cap over the battery terminal after the cable is removed. Note the connection order, and be careful not to make contact between the positive lead of the battery and either the rack or a negative lead.
4. Remove each used battery from the rack.
5. Install each new battery into the rack.
6. Individually connect each battery, starting with the battery farthest from the UPS and moving forward. Be careful not to touch any positive terminal or cable with any negative terminal, cable, or the rack itself.
7. Once you’ve connected all of the individual cables, reconnect the battery racks to the UPS.
8. Verify that the UPS sees valid batteries attached and that it doesn’t indicate any kind of an error.

Adding New UPS Batteries

Adding new batteries to a UPS is very similar to replacing the batteries in a UPS. However, there are some very important differences. The first difference is that not every UPS supports extra batteries. In these cases, you are limited to the batteries the UPS already has.

The second difference involves the need to make sure that the battery cables can handle the current they will carry. When you’re simply replacing batteries, you know that the connecting cables are sufficient to carry the current from the batteries to the inverter. When you’re adding batteries to an existing configuration, however, you need to verify that the battery cables connecting the batteries to the inverter are sufficient to carry the current.

There are guidelines for the size of cable required for a given voltage and amperage, but the best idea is to contact the UPS manufacturer and verify that the cables are sufficiently sized for the additional batteries.

Working with Matrixed UPSs

Somewhere in the middle ground between mid-sized and large UPSs exists a hybrid UPS that allows you to replace modules without taking down the entire system. These matrixed systems make expansion easy by allowing you to add inverter modules to support more load and battery modules to allow the load to be carried longer.

The fundamental difference between a matrixed UPS and a mid-sized UPS with replaceable batteries is that in a matrixed UPS the power generation and distribution components can be replaced independently. This means that it automatically supports a built-in fail-safe bypass mode of operation when the inverter is nonfunctional. In other words, the power will continue to be conditioned by the transformers and MOVs even if the inverter is not present.

This ability is certainly an advantage, but the sales of matrixed UPSs have not been significant, so similarly sized UPSs with replaceable batteries are usually a better deal than matrixed UPSs. (Due to economies of scale, devices that are sold in quantity are generally cheaper than those that cannot be made in similar quantities. This leads to matrixed UPSs being more expensive.)

Replacing an Entire UPS

With smaller to mid-sized UPSs, you’re likely to just add a new UPS to handle new servers. However, if you have a large UPS, you’ll generally replace it with a new, larger unit when you need to support greater loads. Using a single large UPS means that you have fewer units to monitor and maintain.

Replacing a large, hardwired UPS with a new one is a simple process. (Most large UPSs are hardwired into the electrical system of the building rather than being plugged in.) The new wiring is run for the new UPS, the supported systems are shut down, the old unit is shut down, the new unit is connected, and the new unit is started. The biggest catch to this whole procedure is the new wiring. Because the new UPS will support greater loads, it will require more power to run through it. As a result, the wires connecting the UPS to the outside utility and to the internal distribution panel often need to be upgraded. In addition, the internal distribution panel itself may need an upgrade. Take these electrical changes into account when you plan for the new UPS.

Lesson Summary

In this lesson, you learned the basic procedures for replacing batteries in mid-sized and large UPSs, as well as for adding batteries to a UPS. The lesson also discussed issues related to replacing a UPS in its entirety.

Lesson 2: Monitoring the System

Chapter 2, Lesson 4 provided a brief overview of system monitoring and the technologies used to monitor a system. This lesson revisits that topic, focusing on software and hardware specifically designed for system monitoring.

• Identify system monitoring tools and hardware
• Select appropriate tools for your environment

Estimated lesson time: 20 minutes

System Monitoring Hardware

System monitoring hardware comes in two basic categories: hardware that monitors a single server and hardware that monitors the environment in which the server is operating. I’ll discuss the server monitoring hardware first.

Server Monitoring Hardware

No matter how reliable a server’s hardware or operating system is, something will eventually fail. It’s a simple fact of life. Server monitoring hardware is specifically designed to monitor the system and, if there are problems, provide notification and diagnostic tools to get the server back online quicker. The advantage that hardware-based tools have over software-based tools is that they can function regardless of whether the operating system is functioning.

Hardware-based tools provide their own processor, memory, and battery backup so that they can continue working even after the main processor and software have locked up. The battery backup on this hardware generally provides about 30 minutes of operation should the server power fail. This is generally enough time to send notifications to a numeric pager or to send SNMP traps to monitoring stations to ensure that someone is notified of the problem.

Intel has begun integrating server management into its motherboards, and other vendors, such as Compaq, HP, and Dell, offer add-on cards that provide enhanced functionality. Intel also offers a remote management card for non-Intel motherboards and for Intel motherboards produced before the integrated management tools were included on-board.

These add-on cards offer features that aren’t possible in software, such as the ability to monitor failure events, provide outbound paging notification, add to system event logs while the operating system is not functional, and reboot the server. Having a hardware card that is able to create an event that can be set up as an SNMP trap eliminates the need to have a management console check the server continually to determine its status. It also decreases the amount of time necessary to determine that a problem has occurred. Because the SNMP trap is sent almost immediately to the management console, there’s no delay due to the wait between polling intervals. Additionally, the server isn’t burdened with having to respond to the SNMP management console queries associated with continual checking.

Outbound paging notification generally supports both numeric and alphanumeric pagers. The add-on cards generally support the Telocator Alphanumeric Protocol (TAP), meaning that they support virtually every alphanumeric paging device, including mobile phones with text messaging capabilities. Although the paging capabilities are not as sophisticated as those available in software-based system monitoring applications, they are generally a sufficient backup for sending SNMP traps to a management console.

Another important feature of the server monitoring hardware is its ability to log events that have occurred and to allow access to that information even if the server operating system isn’t functioning. By using information obtained from the motherboard, the system monitor can indicate fan failures, disk failures, memory failures, and so on that may have led to the server’s problems. Some system monitoring tools even allow for hardware-based remote control of the system. This feature allows you to control the server even if the operating system doesn’t load. And finally, sometimes the system monitoring add-on cards allow you to see the last screen before the reboot. This lets you review the NetWare abend or the Microsoft Windows 2000 blue screen to review the diagnostic data.

The final capability that most of the hardware cards offer is the ability to reboot the server remotely. This is particularly useful if the server locks up and fails to restart properly or if the operating system fails and isn’t set to reboot the server automatically.

Hardware-based server monitoring may be expensive (costing between $400 and $1,000 per server), but it’s an incredibly effective way to manage servers remotely or to manage remote servers. These hardware tools can make remote access to a system as effective as local access to the system, providing almost exactly the same level of control.

Environment Monitoring Hardware

If you’ve got server monitoring hardware or software, you’ll know when the server goes down or when it quits providing the services that it should. But you won’t know when there are problems with the climate control or when the fire alarm goes off.

That is where environment monitoring hardware comes in. This hardware is designed specifically to monitor the status of the server’s environment. Although some of the features that an environment monitoring device offers can be handled by an alarm monitoring system, others are unique to the environment monitoring devices.

These devices generally monitor the following environmental conditions:

• AC electrical power. The device monitors its incoming power for failure. If the environmental monitor is connected to a UPS, it can monitor the status of the UPS.
• Temperature. The system monitors the ambient room temperature and notifies you once a given threshold has been reached. If you set the threshold 4 degrees above the set point for the air conditioner, it can notify you of a potential climate control failure.
• Sound levels. The unit can generally listen to the ambient noise in the room and notify you when it exceeds a certain level. This feature is useful to alert you when fire alarms go off or when other alarms in the computer room itself go off. It’s a good indication that something is wrong in the computer room.
• Battery. The device monitors the condition of its own battery. This alert is an internal warning to let you know that the unit will soon fail.

In addition to the internal monitoring that the unit performs by default, these devices often have connections for external inputs. These inputs, often called "dry contacts," are provided so that you can attach external devices and have them monitored by the system. This feature is useful for connecting flood sensors, UPS status indicators, and even motion detectors to detect movement in the server room.

Although server monitors are good for any server that requires remote control, environmental monitors are good primarily for situations in which the environment cannot be monitored continuously by humans. Most environment monitoring devices do not have network connectivity and can call only a small number of numbers, but they are useful because they can be deployed in remote locations or in locations where 24-hour support is not possible. These units can be programmed to call a central monitoring facility, where the support personnel can respond to the monitor. If your organization has no central facility that is monitored 24 hours a day, you can program the environment monitoring hardware to call one or more of the people responsible for maintaining the server room.

System Monitoring Software

Despite the advantages of hardware-based monitoring, it also has one major disadvantage: it’s very expensive when compared to software-based alternatives. Because of the manufacturing costs associated with producing hardware, it is generally much more expensive than a software solution. In addition, most hardware-based solutions require the same software as a software-only solution. Software solutions, on the other hand, can be and often are used without hardware-based monitoring support. Software-based systems have three basic types: standards based, proprietary, and hybrid. The standards-based monitoring systems use the SNMP and DMI standards discussed in Chapter 2, Lesson 4, as well as other standards, such as the Self-Monitoring, Analysis and Reporting Technology (SMART) system used for monitoring hard disk drives. These standards-based monitoring systems generally serve as a warehouse for the performance and availability information collected.

Proprietary solutions gather data directly through operating system calls or through other methods of monitoring that are not standards based. Very few solutions are completely proprietary, but a relatively high number of hybrid software-based monitoring solutions exist, because most software with proprietary monitoring features also supports at least some standards-based monitoring.

The list of software-based monitoring solutions is seemingly endless. They vary from relatively simple utilities that graph performance to complex enterprise systems that monitor every server across an entire international organization. Most of these solutions support multiple levels of notification and notification distribution lists. This flexibility is unique to software-based solutions, and it means that notification of a problem can be routed to the appropriate person no matter what time of the day or night it is. Four basic methods are used to provide notification of a problem:

• Console alert. This method displays an alert message on the monitoring console. It is useful only if you have the monitoring software up and running on the console and you’re sitting there watching it. Console alerts are good for constantly monitored systems but are not practical for most environments.
• E-mail. A Simple Mail Transport Protocol (SMTP) message is sent indicating a problem. This method is useful when you have an e-mail address that’s tied to a portable electronic device such as a mobile phone with text messaging or a pager capable of receiving Internet mail messages. The disadvantage of this notification method is that it requires a lot of things to be functional in order to receive the message, including the Internet connection.
• Numeric paging. This method of paging requires a modem attached to the monitoring console that is used to dial a pager number and then provide a series of numbers. This notification mechanism is very unreliable because there’s no confirmation that the message was sent. You simply have to hope that the timing was set up right so that it worked. The cryptic messages are also often difficult to decipher.
• Alphanumeric paging. This paging method allows for more descriptive alert messages, and it’s more reliable due to the TAP used to communicate the message to the pager. This method also requires a modem attached to the monitoring system. It’s probably the best way to receive alert notifications.

Of course, these notifications have to come from the monitoring mechanisms that the software supports. The following sections review the SNMP and DMI standards, so that you’ll have the basic information you need to select a monitoring package.

SNMP

SNMP is a pervasive, cross-platform standard for network management. It was originally specified in the late 1980s and has been revised several times over the years to include new features. An SNMP environment has two basic components: the SNMP agent and the SNMP management console. The agent is responsible for communicating between the hardware or operating system and the SNMP management console. The SNMP management console performs three basic functions:

• Gets the status of a variable on a host
• Sets the status of a variable on a host
• Receives a trap from a host

From these fundamental building blocks, SNMP allows a management console to monitor the SNMP agent, control its configuration, and receive alerts. The primary purpose of SNMP is to monitor other devices and report problems. These two goals are accomplished via polling and in some cases via establishing traps to notify the management console that something is wrong.

Most of the SNMP activity occurs via polling, in which the management console periodically asks the remote device for the status of a series of variables. The management console then evaluates and sometimes records these variables. This process accomplishes two things: it allows the capture of performance data, and it ensures that the remote device is still connected and operational. Polling is relatively processor intensive for both the device and the management console. The more variables that are monitored and the more frequently they are polled, the greater the load is on both the device and the management console.

Increased overhead is a known effect of polling. The more polling you do and the more frequently you do it, the more resources are consumed. Imagine for a moment that you’re the lone worker in an office that has both a front entryway, where you interact with customers, and a back room, where you do the work. Imagine further that one day you’re on your own. You need to get work done in the back, but you have to take care of customers too. You manage this by working for two minutes in the back and then walking up front to see if anyone is there.

What you’re essentially doing, in continuously going between the back and the front, is polling. In addition to the time it takes you to look out front, you’re spending time walking back and forth. You’re also less efficient when in the back because you can’t concentrate on what you’re doing.

In the same scenario, if you added a bell in the front that customers could ring when they need assistance, you would no longer need to go back and forth just to see whether anyone is there. You could wait until the bell rang and then go take care of the customer. This is known as event-driven or alert-driven demand. It’s very efficient from a processing standpoint.

Like the bell just described, SNMP traps are simply alerts that an event has occurred or a threshold has been exceeded. Traps are useful when a device needs to communicate information about a problem to a management console, but it doesn’t make sense for the management console to continuously monitor the device. For example, a trap might be used when performance monitoring data isn’t necessary. Not every device supports traps, however.

Let’s return to our scenario. What happens if a customer won’t ring the bell, or if you don’t hear the bell when it is rung? Unfortunately, the customer may stand in the front for a long time until you happen to come out for something. This kind of problem occurs with SNMP traps. Sometimes the device can’t send the alert because of internal problems, and sometimes the trap that the device does send doesn’t make it to the management console.

For these reasons, traps are usually used in conjunction with regular polling to ensure that the management console discovers the state of the device at some point, even if the trap doesn’t make it. This integrated method is used because of the failures that can occur with traps, and also to limit the amount of polling that must be done.

Another function of an SNMP management console is the ability to configure the remote device by setting the value of some parameters. This function allows the management console to change the configuration of the device, in the ways that the device has allowed through SNMP. Generally, the types of configuration that can be done via SNMP are relatively simple. This function is designed to facilitate simple changes rather than allowing for complete configuration of the system.

The parameters that can be configured are generally relatively detailed, such as the initial size of the maximum receive unit. This parameter is part of the Point-to-Point Protocol. Allowing a management console to change settings like this one permits it to adapt the configuration to changing conditions.

One of the other reasons for setting a variable is to perform actions on the remote device. For instance, with a device that can be rebooted remotely, the manufacturer might allow you to set an SNMP value (a variable) that specifies the number of seconds until reboot. This variable doesn’t require that SNMP support a new reboot command; instead, the variable can use the standard facility built into SNMP to change the configuration of a device. However, it does allow other activities, such as a remote reboot, to be accomplished.

We haven’t talked yet about how messages are communicated between the management console and the device. This communication occurs via a User Datagram Protocol (UDP) port. UDP is part of the TCP/IP protocol family and provides datagram, or nonguaranteed, delivery. That is, the protocol doesn’t ensure that messages aren’t lost or duplicated by the network. The software using the protocol must assume this responsibility.

This means that the SNMP management console must accept the possibility that its requests may get lost and will need to be resent. It also means that the SNMP management console may never see an SNMP trap sent by a device because it may be accidentally discarded by the network. This is one of the reasons that all trap-based monitoring should be backed up with at least some limited polling.

You now know how SNMP communication occurs. But how do the management console and the device know what to talk about? This is one of the most frustrating aspects of using SNMP. Although SNMP has built into it the ability to discover what parameters a device or agent exposes, simply getting the names and values of the remote parameters doesn’t always make their purpose self-evident.

The job of communicating the meaning of the parameters exposed via SNMP is assigned to a file known as a management information base (MIB). This file contains detailed information about the information exposed by a device or agent and a description of what the exposed information means. MIBs are simple text files based on the Abstract Syntax Notation (ASN). They describe the hierarchy that the exposed information takes and indicate whether the information is readable, writable, or both.

For each object exposed by the device, the MIB contains a short section that lists the following:

• Object. The name of the object.
• Syntax. The type of data contained.
• Status. Whether the device must provide the information or has the option of not supporting the entry.
• Access. How the object can be used—for read or for read-write.
• Definition. A text description that explains what the object is for and how it is used. This is the information displayed by the management console when you ask for help about an object.

The importance of these sections and of the ASN format is that they allow you to read the MIB yourself, without loading or installing a management console. Thus, you can review the MIB to determine whether the variables you want to monitor are available from the device or agent.

SNMP is an extremely flexible management framework that allows you to monitor and control a wide variety of devices, including routers, bridges, gateways, switches, and even servers. It should definitely be a part of your overall monitoring solution if your network involves the use of remote communications links, routers, and switches.

DMI

The Desktop Management Task Force (DMTF) developed the Desktop Management Interface (DMI) specification to help address the growing need to be able to control PCs throughout an organization. As mentioned in Chapter 2, DMI and SNMP are similar but different, and they’re not yet interoperable. The DMTF document on the topic of DMI-to-SNMP integration doesn’t paint a good picture for interoperability between the two standards anytime in the future.

One of the fundamental differences between DMI and SNMP is that SNMP is generally implemented on a fixed piece of hardware that doesn’t vary much from one device to another, whereas DMI is focused on PCs, in which hardware is added, removed, and changed frequently. As a result, DMI defines three layers of software instead of the two present in SNMP. They are

• Component interface. Communicates between the hardware and the service provider.
• Service provider. Bundles communication with various components and provides the standardized code for supporting queries and information storage.
• Management interface. Provides a user-level interface to manage and review systems via the DMI architectures.

This three-tier design is more flexible than the two-tier design in use by SNMP, but it is implemented in only a small number of management applications, so it’s not widely used.

Practice: Select an Appropriate Monitoring Solution

In this practice, you’ll review a corporate environment and choose an appropriate monitoring solution.

For the purposes of this practice, let’s examine a firm called Compression Engineering. Compression does rapid product development. It has seven offices throughout the United States. None of these offices have information technology support, but all of them have a Windows 2000 server. The servers are connected together via virtual private networks. The connectivity is relatively stable but has been known to fail from time to time.

The support for the network is handled by two individuals at the central office. Both cover the network in tandem and work well with each other to coordinate activity. Both individuals wear alphanumeric pagers.

What kind of monitoring solution should be implemented?

A hardware-based and software-based solution should be used. Because the remote servers don’t have any support, there needs to be a mechanism to control the server remotely, preferably one that doesn’t depend upon the operating system running. Another reason for hardware-based monitoring is that network connectivity is sometimes not functional, and so it’s important to provide an alternative method for remote support.

Although it would be possible to get by with just hardware-based monitoring, coupling this with software-based monitoring will allow more of the events to be filtered out and, in some cases, handled automatically. This is important because two support personnel supporting seven sites can use all the help they can get.

Lesson Summary

In this lesson, you learned about the different ways in which a system can be monitored, including hardware-based and software-based monitoring. You learned that hardware-based monitoring can be used to monitor servers or the environment that the servers are in.

Next you learned that there are a variety of ways for monitoring software to alert you of a problem, including via direct console messages, e-mail, alphanumeric paging, and numeric paging. You also learned when each of these methods may be appropriate and the problems with each.

Finally, you learned some details of the SNMP and DMI management architectures, including how they work, when they are appropriate, and how they differ.

Lesson 3: Selecting Service Tools

This lesson discusses tools that can help you maintain servers and diagnose problems. Although there are no silver bullets for solving problems, having the right tools with you can make the difference between a day-long project and a simple solution.

• Select the hardware tools to maintain the server and network
• Select the software tools to maintain the server and network

Estimated lesson time: 20 minutes

When selecting hardware tools, you need to decide which ones you want and need. More tools are available than you can efficiently use or afford. This section helps you choose the right tools for your needs.

LAN Cable Scanner

One of the essential tools has nothing to do with the server at all. It’s a device that’s specifically designed to evaluate the performance of a LAN cable, measuring the resistance and capacitance of the cable. In some cases, the cable scanner will be able to review the current network traffic as well as evaluating the cable itself.

Most people don’t realize that one of the things a cable scanner does is to verify that the pairs of the cable are aligned correctly. Ethernet cables use a wiring configuration that splits some of the pairs of the cable, instead of using them side by side. Figure 13.1 shows a wiring configuration for an Ethernet cable.

Figure 13.1 Ethernet wiring configuration, showing a split across pins 3 and 6 (Image Unavailable)

A LAN cable scanner is an important tool because it can help you eliminate the possibility that cable problems are preventing the server from communicating with the clients. It can verify new cables that you create, identify problems that are causing performance issues, and test existing cables.

Multimeter

A multimeter measures voltage, amperage, and resistances. This is one of the lightest and best devices for performing quick checks. You can use a multimeter to check the voltages being output by a power supply. By checking the power of a free drive lead (power connector) like the one shown in Figure 13.2, you can often determine whether a power supply is working properly. The yellow lead should have 12 volts when compared to either of the black wires. The red lead should have 5 volts when compared to either of the black wires.

Figure 13.2 Drive power supply lead (Image Unavailable)

The actual voltage may be plus or minus 10 percent of these values, so the yellow lead may have between 11 and 13 volts, and the red lead may have between 4.5 and 5.5 volts. However, most power supplies maintain voltages much closer than the 10 percent allowed by the specification. Low voltages are often a cause for mysterious lockups. High voltages may be a reason for parts going bad or burning out at a higher than normal rate.

Multimeters can also be used to test the continuity of a cable. This function is useful when you’re installing cables, both internally and externally, and are having trouble with the connected devices but can’t determine why. You can use this tool to ensure that every connector runs through the cable as it should.

ATX Power Supply Tester

An ATX power supply tester has a single, simple purpose: to test an ATX power supply to ensure that it’s working properly. This testing can’t be done with a multimeter if an ATX power supply won’t power up via a motherboard, because the multimeter can’t control the necessary lines in the motherboard connector to instruct the power supply to start to carry the load. This type of power supply tester is generally constructed with some large resistors and an LED. It’s a cheap and effective tool.

POST Card

A power-on self test (POST) card is useful for determining the cause of problems when the system won’t boot. The card is inserted in a free slot in the computer and generally has a few multisegment LED modules that indicate a numeric error code that you can look up in the card’s manual. It’s a quick way to diagnose problems that prevent the system from booting up. Although expensive, this type of card can be valuable if you have many machines that don’t complete the boot cycle when powered on and you’re having difficulty finding the source of the problem.

RS-232 Breakout Box

An RS-232 breakout box is particularly helpful when you’re trying to attach a UPS or other serial device and you’re having trouble getting the communications correct. The breakout box allows you to see the signals on the serial line. It also allows you to reroute signals from one conductor to another until you have a working configuration. Once you have a working configuration, you can make a custom cable to replicate the changes that you made to the breakout box. Figure 13.3 shows a breakout box.

Figure 13.3 Breakout box (Image Unavailable)

Serial Port Tester

A serial port tester is a mini-breakout box that shows the state of only some of the basic signals and doesn’t allow the signals to be remapped to different pins. It’s lightweight and cheap, and it’s often a good way to perform an initial diagnosis of a problem. The device uses dual-state LEDs to show the status of each line, rather than the two separate LEDs that most breakout boxes use.

RJ-45/RJ-11 Crimp Tool

An RJ-45 crimp tool allows you to replace broken RJ-45 connectors on the end of Ethernet cables. The lack of strength in an RJ-45 connector makes this tool a necessity if you don’t maintain a ready supply of replacement cables. The plastic clip that holds an RJ-45 connector in place breaks often, and once broken, the connector won’t stay connected.

RJ-45 crimp tools are cheap and are relatively easy to learn to use, but they can sometimes be frustrating when you’re initially getting used to them. Shoving eight tiny wires into the little space in an RJ-45 connector isn’t easy.

PC Toolkit

It should go without saying, but a standard PC toolkit with a hex driver as well as Phillips and flat-blade screwdrivers should be an essential part of your equipment. These kits may also include other tools, such as soldering irons, that can be useful when you’ve accidentally disconnected a wire or you need to make your own cable.

Software Tools

Hardware-based tools generally bring with them a certain amount of weight. That is, they are heavy to carry around. However, software is very light. In some circumstances you can fit all of the tools that you need on a single CD and a few floppies. Even better, most of the software-based tools you need are built into the operating system and thus don’t require that you carry anything. For that reason, most people have a much larger variety of software-based tools than hardware tools. This section describes some of the software tools you’ll find most useful.

Fsck/VREPAIR/SCANDISK

The first software tool that you should have is the disk repair/scanning utility of the operating system you’re using on the server. You should know how to run it, when to run it, and the options that allow you to perform intensive scans of the disks and volumes attached to the server. Every operating system has some sort of a disk-scanning tool. Sometimes third-party offerings have additional features, such as enhanced file recovery or more in-depth analysis of problems, that may be worth their cost.

Disk Defragmenter

Because servers have many things going on at once and many users trying to write to the volumes at one time, they are more prone to fragmentation of files than a standard system. That is, when the files are written to the volume they are not written in one contiguous string; instead, they are written in a series of small, scattered blocks.

A defragmentation utility rearranges the storage on the drive so that all of the pieces of a file are consolidated in one place, allowing the file to be accessed quicker. Defragmenting can thus give your server a performance boost. A defragmentation run should be scheduled weekly as a part of your preventive maintenance plan. If weekly defragmentation is not practical, you should schedule it as often as is feasible.

Some operating systems, such as Windows 2000, come with disk defragmenting tools built in. Other operating systems, such as NetWare, require the purchase of third-party disk defragmentation software.

Hardware Diagnostics

Hardware diagnostic programs are a great way to exercise a system before installing the operating system or before moving the server into production. These programs can test components sequentially or at random, over a fixed number of loops or until interrupted. The ability of a hardware diagnostic program to repeatedly test hardware components makes it ideal for a burn-in. (Burning in is a process that involves running a computer and exercising it to help identify failures before it is put into service.) These programs do have the limitation that they rarely test add-on cards; they’re normally limited to testing standard components. Most operating systems do not have built-in hardware diagnostics. This is a utility you’ll need to buy.

ARP

TCP/IP uses an Address Resolution Protocol (ARP) to convert IP addresses to media access control (MAC), or hardware, addresses. The ARP utility queries and controls the functions of the cached addresses that the ARP protocol has previously resolved. ARP is included with all server operating systems. Reviewing the ARP cache is useful when you’re trying to determine the hardware addresses of machines with the same IP address, as well as when you’re trying to determine whether the subnet mask and default gateway have been configured correctly.

If you attempt to communicate with a machine that’s not on the local network, the ARP cache should contain the MAC address of the default gateway. This is because the default gateway’s MAC address is the address that the packet should be sent to when the destination network address isn’t on the same network.

ifconfig/BIND/IPConfig/winipcfg

The IP configuration information for UNIX/Linux, Netware, Windows NT/2000, and Windows 95/98 can be retrieved by ifconfig, BIND, IPConfig, and winipcfg, respectively. These commands all tell you the current IP address, subnet mask, and default gateway of the network cards installed in the server. This information is useful for verifying that the DHCP server has provided IP address information to the computer. It can also be helpful when reviewing statically assigned information that you suspect may contain errors.

PING

PING, short for Packet Internet Groper, is a utility that uses the base-level functionality of the TCP/IP protocol suite to verify end-to-end connectivity between two devices. It is built into every server operating system. The command takes one mandatory parameter, the IP address or name of the remote host. PING then attempts to communicate with the other device and indicates success or failure.

Note that when using PING you should try to use the IP address first because this is a more specific connectivity test. By using an Internet name for the device, you’re forcing PING to look up the name at a DNS server before testing the connectivity. It’s possible that PING will report a connectivity problem even if you can communicate with the device, due to DNS errors or problems.

Traceroute

Traceroute is a utility that uses the same basic technique as PING to determine the path that a packet takes through the network. By manipulating a time-to-live field within the TCP/IP packet, traceroute can force routers to progressively report their addresses. This utility is quite useful for identifying where a routing error or disconnection exists. A routing error will result in a loop, which can be seen as the same routers coming up again and again in the same sequence. A disconnection will be seen as a final router reporting that the destination host isn’t available. Traceroute is available as a part of UNIX/Linux and Windows NT/2000. The Windows NT/2000 utility is named tracert.

Nslookup

Nslookup is a DNS lookup utility that allows you to control the type of information retrieved and the servers from which the information is retrieved. This utility helps in the identification of bad name system information that may prevent users from reaching the information they want and need. The redundant nature of the DNS system leads to occasions when some information is correct and other information is incorrect. Depending upon which server responds faster, the user may receive bad information. Nslookup is valid in both Windows NT/2000 and UNIX/Linux, but it isn’t supported on Novell NetWare.

Lesson Summary

In this lesson, you learned about the hardware and software tools that you should at least be familiar with when working on servers. The hardware tools run the gamut from expensive, heavy tools that can be used to solve a variety of problems to small, inexpensive testing tools that have a single purpose.

Software tools are often included with the operating system and are invaluable for troubleshooting connectivity problems. Other tools that aren’t installed with the operating system can be quite useful as well.

Review

Here are some questions to help you determine whether you have learned enough to move on to the next chapter. If you have difficulty answering these questions, please go back and review the material in this chapter before beginning the next chapter. The answers for these questions are located in the appendix, "Questions and Answers."

1. What types of batteries are used in most UPSs?
2. What factor do you need to consider when adding additional batteries to a UPS that you don’t need to worry about when simply replacing batteries that have reached the end of their life span?
3. Why is it advisable to shut down the servers attached to a UPS when replacing the batteries?
4. What is the line-conditioner-only mode of a UPS?
5. What advantages does hardware-based server monitoring have over software-based server monitoring?
6. What conditions in a server’s environment do most environmental monitoring systems monitor?
7. What are the four ways that software-based monitoring systems can notify you of a problem?
8. What are the three functions within SNMP?
9. What are the disadvantages of polling?
10. What are the disadvantages of event-driven or alert-driven notification?
11. How does DMI differ from SNMP?
12. What function does an MIB serve?
13. What problems can a multimeter help you diagnose?
14. What is ARP, and what does it do?
15. Why would someone use nslookup?