Subtitles of the Movie

To conclude our discussion on managing performance and logging, let's talk about maintaining system performance. Now, system performance is something that you actively have to monitor, keep up with, watch your logs, look at your messages, see what's going on with the system and take action when you see something that's causing the system not to perform the way it should be. System performance is a direct factor of several things: memory usage, CPU usage, disk space, network throughput and so forth. The only way to really know if those things are affecting system performance is first of all, when the system crashes or you'd rather this before that, by the way, or monitor it. Monitoring will help you to intercept these performance problems before they become real problems and to monitor trends that may tell you hey, I'm getting for example a lot of paging on the hard drive. Maybe that means a piece of RAM is going bad or maybe that means I need to add more RAM or increase Swap space. So trends like that may alert you to problems that you can fix. Some of the things you can do to enhance your system performance and keep it going is you may require additional RAM, more disk space or faster CPUs or adding more CPUs to a multi-CPU system. This might happen if you've increased the role of the server, if you've added 500 new users and it's a file server, if you've create more services that have to run and provide services to the network. You also may need to offload processing or services. You may have to actually put in another server and divide up the workload between the two servers. Network performance also should be looked at for throughput, bandwidth and so forth to make sure that you're getting the right network throughput that you need for whatever applications are running on the server. You may have to add faster network cards or multiple network cards to offload some of the network throughput problems. You also may have to offload these network services to other hosts. Instead of running file services and DNS and DHCP on the same box, you might want to take those and put two of them on another box, especially if you have a lot of users on the network. You could look at other solutions as well. You could look at server clustering for increase failover capability and of course for performance. That way if something happens to the server, you've got a failover and users may or may not notice the fact that one of the services failed because you've got good failover. Some solutions that you could also throw in there if you needed to, although they're a little bit more expensive solutions are SANs, Storage Area Networks and Network-Attached Storage Solutions. That may help you offload some of the processing power or some of the workload rather to these devices. Let's go ahead and take a look at one more system performance monitor and it's actually a nice, pretty one in a GUI. It's something that's very useful and kind of gives you the big picture all rolled into one. We're looking at the System Monitor in OpenSUSE 11.2 and basically it's one-stop shopping for monitoring the system performance and so forth in Linux and this is very nice, pretty little GUI version. Actually it gives you nice, colorful graphs that are constantly updated and you've also got some of the other facilities here that will help you monitor system performance. Gives you information about the system and available disks pace, processes that are running and what their status are and how much CPU time they're keeping up, resources, which you've already looked at and with resources we look at CPU usage, memory and Swap usage and network bandwidth usage, throughput usage. We can also look at file systems and kind of see what's available and what's used and how much space we left and that might tell us that we need to add additional hard drives. We can also look at hardware information in here and see kind of what drivers are being used and what resources such as IRQs and so forth is being used. So it's kind of a one-stop shopping utility here and while I don't necessarily advocate this utility over any other one, it's really nice to have this all in one place and that's why a lot of people actually like GUIs because sometimes monitoring and managing things at the Command Line can be a little cumbersome and sometimes you really need to see everything in one place visually in order to catch problems as they happen. So this is just one additional system monitoring program that I want to show you and show you how it monitors resources, processes and so forth to detect trends in system performance so you can maintain that system performance that your users are used to having on their system. So we've talked about system performance, we talked about maintaining it and some of the things you need to do to make sure that your performance on your network and on your UNIX box stays at the minimal acceptable level; hopefully better than that.