Previous Table of Contents Next Here are some examples of service monitoring packages, listed from the primitive to the sublime: o Sitter (ftp://feldman.org/pub/sitter) A primitive Linux bash script that checks ports using the Telnet trick and complains to the console when problems are found. This definitely must be manually reconfigured for your site. o Netoscope (http://www.basta.com) A Windows utility for checking socket availability of certain servers. Issues audible alerts or runs a program (such as email) when problems occur. o IPSentry (http://www.ipsentry.com) A socket-monitoring package that runs under Windows. It tries whichever IP addresses and sockets you want, as often as you like. It can also page you when a problem occurs. o WhatsUp and WhatsUp Gold (http://www.ipswitch.com) This is a graphical network polling package with paging features. WhatsUp Gold has Web access, SNMP features, and IPX monitoring, as well. WhatsUp Gold also has the dubious distinction of being quite expensive for a polling package (close to $1,000 at this writing). So when do you use "pollsters" rather than SNMP managers? It depends on what you need. In my experience, only larger shops tend to use the SNMP managers; they tend to be expensive, resource intensive, and complex to install and maintain. Of course, SNMP managers also provide you with lots of information. In particular, the SNMP managers will provide you service-level information-that is, not only will they tell you whether the resource is "up" but also how fast it responds, how many bytes of resources are left, and so on. This makes SNMP managers very valuable if you're looking to optimize your network. SNMP managers are excellent for keeping long-term track (baselining) of how your network and resources look on a normal day. As you'll see in Hour 23, baselining is a very important part of problem determination when, as they say, "The network is slow!" After all, if you don't have a sense of how things typically are, how can you tell when something is out of range? Long-term service-level record keeping is also a way to know-and a way to show your boss-that everything is operating efficiently and is working properly. Still, simple network polling packages can be extraordinarily effective, even for large shops, but particularly for smaller shops. They're typically inexpensive compared to the costs of deploying even a modest SNMP installation, and they're very easy to configure and use. Although most network polling packages don't address service-level issues, look for more of them to do so in the future. Of course, you'll never have the rich level of information that SNMP provides, but you might not need it. In short, a polling package is a one-stop application where you can monitor many services easily. Figure 22.3 shows one of them. [22-03t.jpg] Figure 22.3 Basta's Netoscope just keeps polling away! Real-Life References Network management can be as expensive as you let it be. Of course, ostensibly, when you install a network management station and configure your network for management, you're trying to save time and avoid downtime-both of which contribute to your bottom line. Still, before you go out and plunk down your hard-earned dollar, be sure to go and check references on any expensive network management package. More than that, try to find businesses that operate the way yours does. You can learn from their experience with network management, and you can possibly avoid some expensive mistakes-or be told, "C'mon in, the water's fine!" Either way, you'll know what to expect from your network management experience. Summary "Network management" can be a slippery phrase; it's used to refer to everything but the kitchen sink. In its most helpful form, network management involves network monitoring, which allows you to assess the health of your network. SNMP (Simple Network Management Protocol) is at the heart of complex solutions; many vendors supply reasonably complete MIBs and agents for their equipment and software, which can then be read by SNMP management stations. SNMP management stations are really good at long-term resource trend monitoring and can help you "baseline" your network. SNMP agents can broadcast traps to the network, which are then picked up and acted on by management stations. Typically, notification (such as activating a pager) is the best that most management stations can "manage." SNMP traps can be annoying or they can be a blessing, depending on whether they're waking you up in the middle of the night for a printer jam or to alert you that a public safety system is down. Generic socket-polling packages for TCP services are also available, and so are service-specific polling packages. The generic packages are limited in that they typically only tell you whether a service is "down" or "up" (that is, they're pass/fail graders). They generally don't report how well the service is doing on resources and response times. Some of the service-specific and vendor-supplied monitoring utilities, such as the monitoring applications that come with certain servers, switches, and routers, will give you an SNMP-like wealth of information without the overhead that SNMP implies. Depending on the amount of money you're going to spend on a network monitoring solution, you'll want to do some legwork and check references thoroughly before you invest. Although any polling software solution is typically cheap enough to be throw-away if it doesn't meet your needs, a complete SNMP solution is expensive enough in time and money invested that investigating how well it does in a similar shop is a good idea. Previous Table of Contents Next