Jump to: navigation, search

Resource Planning for Application Servers

The two major aspects that affect resource utilization are the selected log file collection method and log file generation rate. A careful analysis of these two aspects is highly recommended.

Collection Method

There are two options to transfer copies of the application log file to Central Log Storage (CLS):

  1. Collect log files on local storage and start the log file transferring job during off peak-hours: This option requires large local storage, and the existence of off-peak hours. It also introduces some extra load on the file system, delays in log file availability for processing by CLS services, and the risk of losing essential information in case of a system crash.
  2. Transferring log files in quasi-real time: The log file transferring job is started periodically and log files are transferred to the CLS at more frequent intervals. This option introduces extra load on CPU, I/O and network utilization. This option requires extra network capacity for its operation.

Log File Generation Rate

From the LFMT perspective, log file generation rate is the main parameter that defines performance impact on CPU, I/O and network utilization on the Genesys application servers. A higher rate of generation creates more of an impact on an environment. The log generation rate can be calculated by the following formula:

LogSizePerSec = CallRate * LogSizePerCall

Where LogSizePerCall is based on observation/testing of the Genesys environment.

Network Bandwidth Utilization

Network bandwidth required for LFMT needs to be provisioned carefully. If it is under-estimated, the log files transfer could consume significant bandwidth and impact the response time of the Genesys application servers. Network bandwidth utilization by the LFMT can be calculated by the formula:

NetBandwidth = ( LogSizePerCall * CallRate ) / CompressionRate

Where CompressionRate = 10 to 12 (depending on compression performance).

Some recommendations:

  • Provision some reserved network capacity (bandwidth).
  • Carefully select the log transferring job interval to prevent system overload.
Important
Precautions must be taken when configuring scheduled log collection intervals. One iteration of a task should be completed before another iteration starts.
  • Consider using the throttling capabilities of LFMT to preserve network capacity for other system components.

These are some examples of network bandwidth calculations:

For CallRate = 5 calls/sec, NetBandwidth = 1.2 * 5 / 10 = 0.6 MB/sec.

If max network transfer rate is set to 1.2 MB/sec and interval= 5 mins, then the log file transfer will be finished in about 2.5 mins.

For CallRate = 10 calls/sec, NetBandwidth = 1.2 * 10 / 10 = 1.2 MB/sec.

Disk Storage

Use the following formula to calculate the storage capacity required to store log files on the Central Log Storage.

TotalLogSize = LogSizePerCall * NumberOfCalls

Where NumberOfCalls is the number of calls for which the log files are stored on local storage before moving them to CLS.

CPU

The LFMT log collection process produces approximately 5-15% overhead when packaging and sending files. Therefore, it is recommended to monitor the Genesys application servers to ensure that the CPU is capable of handling up to 15% additional load during peak hours.

Memory

The LFMT packaging process uses approximately 100MB of additional memory.

Logging Levels

The logging level affects the performance of the application and the amount of data that needs to be stored and transferred over the network. Therefore, it is recommended to follow Genesys best practices to set up the appropriate logging levels for each of the Genesys applications running in your environment.

This page was last modified on March 1, 2018, at 09:05.

Feedback

Comment on this article:

blog comments powered by Disqus