Object virtualization with surrogate

Question

question

Roi Sánchez asked Apr 24, '20 Phil BoBo edited Apr 24, '20

Object virtualization with surrogate

Hi there,

We are working on a warehousing model in which performance is critical because of flow items quantity and logic difficulty. Thus, we need to virtualize items in the storage system. Some of these items have surrogate objects attached (surrogates because we think it is better than flowitems subnodes in terms of performance).

The thing is when virtualizing and recreating them, the surrogate object seems to be missed and replaced by another one (we think it is the first one that entered the storage system). Is this a bug or there is something that we are doing wrong?

We have reproduced this issue in a dummy model that you can find attached.

Object virtualization with surrogate.fsm

Thanks!

Roi.

Software Version:

FlexSim 20.1.0

flexsim 20.1.0

object-virtualization-with-surrogate.fsm (48.5 KiB)

Answer 1 · 2020-04-24T15:53:17Z

Phil BoBo answered Apr 24, '20 Phil BoBo edited Apr 24, '20

When virtualizing items, the items themselves are destroyed. Only the first item is preserved and duplicated when the items are un-virtualized.

Any pertinent information should be kept on labels specified on the Items tab of the Storage System Properties.

This is by design. These limitations are on purpose in order to force you to structure your model to be performant.

See the Virtual Items section of https://docs.flexsim.com/en/20.1/Reference/Tools/StorageSystem/

"Because only minimal information is kept, virtualizing items can save significant memory. In addition, virtual items are also able to draw significantly faster. The drawback is that the flowitem is destroyed, which can make some models a little more difficult to create."

Although "a little more difficult," you need to change your model to account for how virtualization works if you want to take advantage of its performance improvements.

I don't know why you think storing items in the drawsurrogate has better performance than subnodes, but that's a bad assumption that's just complicating your model.

question