In a previous job, I developed and supported an application that performed ground water data management and analysis. I also answered the phone when customers called with questions. The following is a real conversation that I had with a customer (somewhat edited for space).
Me: “Hello, this is software support. How can I help you?”
Customer (in a not so pleasant tone of voice): “Where is the data management button I requested in the last release of the software? I should be able to press one button and all my problems should be solved!”
Me: “Thank you for your inquiry. There is no one data management button, but if you do the following…”
Customer (a few decibels lower): “Oh…ok…so how do you do that again?”
Wouldn’t it be great if software engineers could design a single button to perform all the functions we needed or wanted in a system? You might ask the same question when working with IBM PureApplication System. Where is that monitoring button that will allow me to look at my compute nodes, middleware, system status, application servers and networking devices? The answer is there is no single button; however there are many individual “buttons” that you can “press” to monitor your PureApplication System. IBM software designers and engineers have done a fantastic job integrating several monitoring packages into IBM PureApplication System.
PureApplication System is a hardware and software intensive product, so the built-in monitoring facilities are quite extensive and detailed. In a previous blog post, I outlined how you can use the hardware infrastructure map to aid in the diagnosis of compute node problems. From the PureApplication System console, you can monitor the hardware infrastructure, compute nodes, management nodes, storage devices, and network devices. The following examples are just a few of the monitoring “buttons” that you can “press’ from PureApplication System’s workload and system consoles.
The number of compute nodes is dependent on the size of the system. This system has three compute nodes spread across three cloud groups. Each compute node has a backup for redundancy. From this view, you can monitor in real time each physical core, physical memory of the compute node, and each individual virtual machine that has been deployed to the cloud group assigned to the compute node. There is quite a bit of information packed into one window.
You can monitor the PureSystems Manager (psm) or the Virtualization System Manager (vsm). The information displayed for each manager is identical. You can monitor information such as the number of jobs running on the node and the total amount of physical memory used. The difference between the two nodes is in their functions. Among many others, the psm manages resources such as the system catalog, virtual systems, and virtual applications. The vsm manages the actual virtual machines and associated resources.
This view shows you information about the actual storage node and storage node expansion. The configuration shown in this example has two storage nodes with identical storage node expansion for each node. You can monitor the storage capacity from here and make deployment/management decisions. You can monitor the storage volumes and storage controllers as PureApplication System spins up virtual machines and allocates capacity.
As a refresher, the two top of rack switches are used to connect to the external network and are configured by the customer. The SAN switches and network switches are preconfigured at the IBM factory and are not configurable by the customer. These are used for internal communication with in the rack. With that in mind, let’s take a look at one of the top of rack switches. Remember that these switches should be configured identically for failover and redundancy. The customer ports include all ports that connect to the external network. If you were to expand this field, you would see status on state, speed, input packets, and output packets for ports 41 through 56, and port 64. Expanding network ports would show the same information for ports used for internal communication within the rack.
Virtual Systems (specifically WebSphere Application Server)
If you are familiar with WebSphere Application server, you will be pleased to know that you can monitor your WebSphere environment just as if it were in a non-virtualized environment. In this example, a small cell was deployed. The cell consists of a deployment manager and a custom node. Here is the cell as shown in the list of virtual system instance.
The cell is represented by two virtual machines. One represents the deployment manager; the other is the custom node.
If you expand the deployment manager and scroll all the way to the bottom of the information for the virtual machine, you will find a link to the WebSphere console.
Following the link allows you to log in to the console. Depending on the access rights of the user, you can use the console to manage your WebSphere environment just as you would in a non-PureApplication System environment.
In summary, there are numerous monitoring “buttons” that allow you to monitor hardware, middleware, and the overall system. Most monitoring information is easily accessible from either the Workload console or the System console.