The Safe Software Blog
Author:
Gerhard Fischl

Google
Get the Blog Newsletter

Delivered by FeedBurner

About FME    |   April 25, 2018   |   By Gerhard Fischl

FME Cloud FAQ: Enabling unlimited FME Engines for your workflows


There are a few reasons you might choose FME Cloud over FME Server. If your workspace connects to data that resides in the cloud, like web APIs, data stored in a cloud drive, or satellite imagery, then deploying your FME Server instance in the cloud will make the workflow run a lot faster. For many users, the cost of FME Cloud is also much lower, because you only pay for the time your server spends running. FME Cloud further removes the IT hassle of hardware and installation.

Let’s look at how FME Cloud can also be advantageous by letting you run multiple FME Server Engines on one processing core.

On FME Cloud, you can run as many FME Server Engines as you want at no extra cost.

You pay for what you use on FME Cloud. You can decide between on-demand pricing for hourly charges or go with a yearly subscription (cheaper for instances that run more than 16 hours a day). The price is determined by the size of your instance and your usage. Once the instance is running, you can define the number of FME Server Engines you want based on your workflows—the larger the instance, the more Engines you will be able to sensibly run.

Multiple Engines? What about the “one Engine per core” rule?

When we discuss solutions with FME Server, we often recommend one FME Engine per processing core running on a production machine. This is because a single FME Server job running on a single Engine can potentially utilize one core by 100%. To make sure everybody can take advantage of the the full potential of FME, we recommend planning for this scenario.

On FME Cloud, this recommendation is the same if you run CPU-intensive jobs (e.g. raster or point cloud processing). But if you run a lot of small jobs (e.g. connecting APIs or cloud services) you might be able to run 10 Engines or more on 2 cores without putting too much load on the server. The difference on FME Cloud is that it’s easy to experiment with your instances. You get great monitoring tools out of the box that will allow you to quickly find the right amount of Engines for your FME Cloud instance.

How do I decide how many Engines to run?

After you launch your FME Cloud instance (get started with a $250 FME Cloud credit) and run a couple of jobs, you can increase the FME Server Engine count via the FME Cloud Web UI. Pay close attention to the FME Cloud Metrics to see if your instance can handle the Engines you started. The most important metric is the server load. As soon as you reach 100% utilization of your processor cores, your instance will slow down significantly. Ideal utilization is about 75% to allow for some headroom, but depending on your workflows it might make sense to push close to 100%. After all, it’s easy to set up FME Cloud alerts so you can find out if your instance is in trouble.

*

Read more about FME Cloud: