[pmiguel]

It may mean the size of your DNA before sequencing. By observing the sizes of DNAs, we can check contaminants. And they will fragment DNAs and sequence them. Later you will get the sequencing data.
By the way, your data looks strange.
And this result does not have nothing with Illumina library insertion data.
Generally, the insertion size can be 180-350 bp.
You better BLAST both sequences of the same id and manually check the insertion size.

What is strange about it?

The 35bp and ~10.3kb peaks, are the spike-in size standards that Bioanalyzers use for DNA high sensitivity chips.

Normally TruSeq "PCR-Free" libraries will produce a range of product sizes similar to this. Of course, the longer products will fail to produce sequence-able clusters. They seem to be unable to compete with the shorter products.

Of course the pure-programmers probably aren't reading this sub-forum. But to any who are: Please don't require insert sizes to run your assembler/mappers! If your algorithms really need that information, figure it out! The bioanalyzer results don't give you an accurate assessment of the insert sizes, for the reasons I describe above. So asking the lab what the average size of the inserts were is not terribly useful!

--
Phillip