Performance considerations

In order to get the most out of OL Connect, it is important to determine how best to maximize performance. The following guidelines will be helpful in extracting the best performance from OL Connect and they give a rough indication when it would be useful to start looking into hardware upgrades.

Performance analysis details

OL Connect's output speed is limited to a certain number of output items (web pages, emails, or printed pages) per minute. What the maximum total output speed will be is determined by your license (see Speed quota: Pages Per Minute).

To get an indication of the actual Print output speed, output a Print template to a AFP (Enterprise edition only), PDF or PS file, using one single Weaver engine and the maximum target speed per job (see Parallel Processing preferences).
Next, open the log file of the Weaver engine. By default, the log files are located in this folder: C:\Users\[username]\Connect\logs\WeaverEngine, where [username] is your own Windows user name.
Search the log file for "PPM" (pages per minute). Repeat this a few times to determine the average output speed.
Likewise, the output speed for an Email or Web template can be found by running it with one Merge engine and the maximum target speed per job. In the Merge engine's log file, search for "PPM".

If your jobs are not running at the licensed speed, there may be several ways to improve performance, as described below. Make sure to address all issues mentioned in this topic before deciding that you need to invest. Note however that it is not guaranteed that the licensed speed can be achieved with any job. Creating output for templates with very complex scripts or complex graphics resources will take a certain amount of time, even on high-end hardware.

Improving performance beyond what can possibly be reached by using the methods described below requires purchasing a Performance Pack (see Performance Packs).

Improving performance beyond what can possibly be reached by using the methods described below requires purchasing either a Performance Pack (see Performance Packs) or upgrading to PReS Connect.

For advice please contact your local sales office (see Upland Objectif Lune's Contact page).

Engine configuration

As explained in another topic (Connect: a peek under the hood) the Connect Server cooperates with different engines to handle specific tasks.

A DataMapper engine extracts data from a data file. A Merge engine merges the template and the data to create Email and Web output, or to create an intermediary file for Printed output. The intermediary file is in turn used by a Weaver engine to prepare the Print output. A Merge engine merges the template and the data to create an intermediary file for Printed output. The intermediary file is in turn used by a Weaver engine to prepare the Print output.

Configuring these engines to match both the hardware configuration and the typical usage situation is probably the most effective way to improve Connect's performance.

The number of engines is one of the "Parallel Processing Preferences" that let the Connect Server manage its workload in such a way that the highest possible output speed is achieved.

For an explanation and guidelines to these settings, see Engine configuration and Parallel Processing preferences.

Note: Connect Server and Connect Designer each have their own distinct scheduling preferences.

Use the Connect Server Configuration tool to change the Connect Server settings and Designer > Windows >Preferences for changing Designer settings.

Template optimization

When you find that the speed per Merge engine - the Content Creation speed - is low, optimizing a template can make a huge difference. For advice on how to optimize a template see: Optimizing a template.

Data optimization

One record that produces thousands of pages will take long because only one merge engine can work on it. To produce output faster, try splitting the data into multiple records.

Network and internet connections

Use a fast network and internet connection or avoid loading external or internet resources. Using images, JavaScript or CSS resources located on a slow network or on a slow internet connection will obviously lead to a loss of speed. While we do our best for caching, a document with 5,000 records which queries a page that takes 1 second to return a different image each time will, naturally, slow output generation down by up to 83 minutes.

Hardware configuration

When processing speed is important, the following is suggested after addressing the other issues mentioned in this topic.

  • Antivirus exclusions. Sometimes, virus scanners, other security software or indexing services can interfere. It can help to disable those kinds of tools for the areas where Connect stores intermediate files. You could exclude the entire C:\Users\<connectuser>\Connect folder. See also: Antivirus Exclusions.
  • Use a high-performance, low-latency hard drive. Connect benefits from fast I/O. This is especially true for DataMapper engines (see DataMapper engine). Preferably use a Solid State Drive (SSD) or similar for storage.
  • Use at least 8+ GB High-Quality RAM. Check memory usage while the Print command is being executed to see if you need more than the minimum of 8GB. Assuming that the Connect Server and the Connect database need 1GB each, and that each engine needs 1GB as well, you can roughly estimate how much memory is needed.
  • Consider using a physical machine instead of a virtual machine. When running on a Virtual Machine, the machine may report that it has sufficient hardware (cores) available, but in a virtual environment you need to make sure that this hardware is not being shared with lots of other virtual machines.
  • Consider using hardware with more physical cores. OL Connect doesn't limit the number of Merge engines that is used for a Print job, so if the number of physical cores is low, it makes sense to see if that can be increased. When running on a virtual machine, this is usually easy. When running on a physical machine, it means that you may have to switch hardware.
  • For both virtual and non-virtual environments, make sure the machine is not busy with all kinds of other processes.