Web Server Capacity Planning

Server capacity planning is the process of using your understanding of current server capacity usage to optimize present and plan for future server performance. At its core, server capacity planning involves monitoring and alerting on server resources, such as CPU, memory, disk, and network. Being able to forecast when your server will run out of resources can also help you optimize hardware investments, while ensuring you continue to provide the capacity required by your applications.

Server capacity planning involves monitoring server capacity and resource usage to understand current use and how it may change in the future. This process helps you find which server resources have reached or are reaching capacity, so you can make more informed resource investments.

The first step of server capacity planning is analyzing how current capacity is or isn’t meeting the needs of users by monitoring and measuring server CPU usage, server memory usage, virtual memory capacity, and other server resources with an impact on response times. You can use this information to optimize current capacity by finding and limiting users or activities that strain specific resources.

Using information about current server capacity in combination with forecasts about future business activity, potential user growth, and seasonal changes in demand, you can begin estimating and planning future capacity needs. A modeling tool or server capacity calculator can also help you more easily determine future needs.

When it comes to SQL Server capacity planning, the general process is largely the same as general server capacity planning. The core element of capacity planning in SQL Server is knowing how much data is currently stored in the server, the number of users on the system, and what it means for your applications and number and size of transactions.

Getting these calculations often involves monitoring transactions in SQL Server, which you can use to begin planning future SQL Server capacity needs and forecast future growth.

While storage capacity planning tools focus primarily on making sure you have sufficient disk space, server capacity planning tools provide key historical and current resources and their utilization provided by the server to help you ensure optimal speeds and quality of transactions between servers and applications. If your resources aren’t being optimally used, you can experience bottlenecks and reductions in service delivery.

Server capacity planning tools are built to monitor consumption of CPU, storage hardware, and memory to see if resources are available, or what’s at the root of server resource contention. Once you identify what’s responsible for overconsumption, you can use the tool to end processes and better allocate resources across virtual and physical servers. The same goes for when you’re underutilizing resources.

SolarWinds Server & Application Monitor is a server capacity planning tool designed to help you optimize resource use to support your present and future capacity needs.

SAM tracks critical server resources including CPU, memory, I/O, and disk to help you find what processes are causing contention for resources, so you can kill these processes to more easily right-size resource allocation.

SAM can also help you plan for future server capacity needs by collecting capacity trends over time, SAM provides capacity forecasting charts to help you quickly understand how much capacity you have and when you’re likely to run out of a specific resource. SAM simplifies forecasting usage and sees when server resources are approaching warning and critical thresholds. This way, you can find and fix server capacity problems before they affect end-user experience.

Monitoring the capacity of your server clusters is simple if you have a tool to keep track of cluster health on a Microsoft Windows Server failover cluster. With SolarWinds Server & Application Monitor, that’s exactly what you get.

To set up cluster health monitoring, follow these steps:

• Log in to your web console as an administrator. In SAM, you need to log in to the Orion Web Console.
• Hover your mouse over the settings and select Manage Nodes.
• Add a new node with the VIP of the cluster as the node. Choose “External Node” as the polling method.
• Then click Settings and navigate to All Settings.
• Under the option Product Specific Settings, click on SAM Settings.
• Select Manage Templates.
• Finally, apply a Microsoft Windows Server XXXX failover cluster (in this case, XXXX is your Windows Server version).

You can follow the same basic process to monitor your SQL Server clusters capacity by adding the VIP in the Orion Platform. Once your clusters are being monitored, the tool will automatically monitor the resources contained in the cluster resource group, regularly polling them, and putting relevant results in your dashboard. This monitoring can help with servers capacity planning since it helps you understand your server resource needs and helps you reallocate those resources as needed—before a server runs out of capacity and the only option is to throw money at the problem.

SAM includes templates to help with effective server cluster monitoring. These include templates for:

• Microsoft Windows Server 2003 Failover Cluster
• Microsoft Windows Server 2008 Failover Cluster
• Microsoft Windows Server 2008 R2 — 2016 Failover Cluster (Advanced)
• Microsoft Windows Server 2012 — 2016 Failover Cluster
• Microsoft Windows Server 2012 Failover Cluster

These templates can be adjusted to fit your needs, including when it comes to setting baselines for your server sizing calculations.

Setting baselines to make server sizing calculations with SolarWinds Server & Application Monitor is designed to be simple. Generally, SAM calculates baselines on demand whenever you want them. However, it’s recommended you include at least seven days of data for your baseline data calculations to be considered accurate.

Once SAM calculates your baseline thresholds and applies them to your component monitors, those thresholds will remain static until you manually reapply them. They won’t move each day based on new calculations for the previous seven days of data, since a moving baseline could obscure anomalies like data spikes that really need to be highlighted to maintain performance.

SAM uses three macros to calculate baselines. These are:

1. ${USE_BASELINE}: This calculates the baseline currently being used in SAM. This is the value being used in the threshold fields for SAM component monitors.
2. ${MEAN}: This calculates the current Average or Mean.
3. ${STD_DEV): This calculates the Standard Deviation.

While Server & Application Monitor is built to do many of these calculations automatically, the tool lets you adjust to help ensure your baselines and server sizing calculations meet your specific needs. Some of these adjustments include the following:

Changing the amount of data used in baseline calculations

• Navigate from Settings, to All Settings, then to SAM Settings, and then Data & Database Settings.
• Enter the number of days’ worth of data you want included.
• Click Submit.

Applying baseline thresholds at the template level

• Navigate from Settings, to All Settings, then to SAM Settings, and then to Manage Templates.
• Select your template and click Edit in the toolbar.
• If you’re selecting just one component monitor, you can click [+] to expand the monitor details. From there, you can click on Latest Baseline Thresholds to show the ${USE_BASELINE} macro in the Warning and Critical fields. Select the options for sustained thresholds where you need them.
• If you’re selecting more than one component monitor, you can check the boxes next to the names of all the monitors you want to deploy. Then you can click Multi-Edit and review the Statistic Threshold data in the pop-up window. Make sure to check the Use thresholds calculated from baseline data.

In addition to these changes, SAM also lets you apply baseline thresholds at the application level, use the latest baselines for a component monitor, and edit capacity planning thresholds for a node.