LAE

Note: This was originally posted by an inactive account. Content was preserved by moving under an admin account.

Originally posted by: dglavastorm

Hi, not actually sure if I've asked this question previous but can't find it if I have.

I've got 32 bit Desktop installed and wondering if it's worth the hassel to change to a 64 bit.

Is there a functional difference between 32 and 64 bit desktop Lavastorm ? Is 64 faster, can it handle bigger data sets etc . ?

Cheers,
Dave

Note: This was originally posted by an inactive account. Content was preserved by moving under an admin account.

Originally posted by: awilliams1024

Hi Dave,

It really depends on how you are using LAE.

The LAE Desktop product runs on both 32-bit and 64-bit Windows desktop OS platforms. The LAE Desktop edition software installer includes both the BRE client application (used to create data flows and initiate their execution) and the LAE Server component (which executes the required data flow locally on the user's PC). Both are deployed to the local machine when the installer is run.

Regardless of whether you are using the 32-bit or 64-bit Desktop edition, the BRE client itself is a 32-bit Windows application. However, the LAE Server component is different for 32-bit and 64-bit platforms - for instance, a 64-bit JRE is used with the latter version.

If you are using the BRE client to execute a data flow on a separate (Windows Server or Linux) LAE Server machine/ Server farm then there is not going to be much difference in practice since the data flow execution is being done remotely.

If you are executing data flows locally on your PC then there will be benefits of using a more modern architecture (and its probably going to be a faster machine anyway, given Moore's law). The movement of data into/out of memory will be faster and a high IO capacity is important for good execution speed. The Java-based nodes will be using the 64-bit JVM and may be more efficient (though it depends to some extent on the 'shape' of your data and the calculations you are performing on the data). Using a 64-bit JVM means that the system can be configured to start the JVM with a larger heap size, if required.

The majority of nodes in LAE do not run 'in-memory' as records are processed sequentially rather than as an entire data frame. There are a few exceptions to this such as when using the R language statistical and predictive nodes and the Lookup node.

Each node runs in its own JVM so having a 64-bit machine means that when multiple nodes are being run in parallel they can utilize more memory than would have been available on a 32-bit machine.

You also need to consider whether you are accessing external systems such as Oracle or Teradata as you typically cannot mix 32-bit and 64-bit drivers.

Regards,
Adrian

Note: This was originally posted by an inactive account. Content was preserved by moving under an admin account.

Originally posted by: dglavastorm

Cheers Adrian, a very detailed but easy to understand answer :-) Will look at upgrading.