server monitor.png

Are your Microsoft Servers running slow?  Read this post for best practices on the top metrics to monitor.

 

• When experiencing an application or server performance issue in your environment, perhaps the most obvious metric you’re likely to look to first is the affected servers CPU utilization. This metric provides insight into how much load is being placed on the servers’ processor at any given time. A high and sustained CPU utilization may be indicative of underperforming hardware that may need replacement or upgrade. If the server is virtual it may suggest that the virtual machine suffers from insufficient resource allocation. If the machine provides multiple services and functions, you may also want to consider distributing those roles amongst other servers in your environment to distribute the load more evenly.

 

• Another likely culprit of poorly performing applications and sluggish servers is the machines physical memory consumption. RAM is where the operating system stores information it’s actively using to service actively running applications running on the host. When a server has an inadequate amount of memory to run both the operating system and the applications that run on it, the OS will begin moving lesser used blocks of memory temporarily to virtual memory located on disk. This commonly referred to as paging. As demand for memory resources increase the more paging occurs. Because the disks are significantly slower than RAM this introduces a bottleneck on the server that can significantly impact overall server performance. Should this condition occur for a prolonged period of time you should consider adding additional RAM to the physical or virtual server.

 

• As virtual memory consumption increases hundreds of megabytes of information are constantly moving from RAM to disk and back to RAM again. This puts tremendous strain on the physical disks where the swap file is located. It’s always best to ensure your operating systems swap file is located on a different drive than the operating system to prevent swap file fragmentation and to ensure paging doesn’t impact other disk I/O intensive operations such as databases.

 

Disk performance is the actually the leading cause of server and application performance issues today. Big data and virtualization have only compounded this problem by placing ever increasing additional strain on servers’ disk I/O subsystems. As such, it’s important to keep close tabs on your server’s queued I/O and disk latency to understand how your storage performance is impacting your applications. When either disk latency exceeds 100ms for any period of time this is likely indicative of a storage performance issue. The same can be said of sustained high disk Queue length. If your server is suffering from poor storage I/O performance consider changing your RAID type, adding more physical disks to your array, upgrading your storage controller to one with larger cache, or replacing older, slower disks with solid state or 15k SAS drives. Alternatively you may be able to more evenly distribute your applications disk I/O load by moving databases, applications, temp files, etc. across multiple disks.

 

• Finally, server monitoring should include keeping an eye on the hardware of your Windows servers. If there is an underlying problem with the hardware, the application may not function correctly, and an unforeseen hardware failure (hard drive, fan, etc.) can take your application down without any warning.

 

Learn how to monitor and manage other aspects of your Microsoft environment – www.solarwinds.com/gotmicrosoft

by Jennifer Kuvlesky