The closer your match generationally you will generally have more shared centimorgans (cMs).
There is a published shared cM relationship range chart on the ISOGG website that gives a good general 'guideline' for interpretation..
...but it is just a guideline as the further one gets out in relationship you may retrieve additional shared cMs from more than one shared ancestor or path to a common ancestor.
Though the general rule is you inherit about 50% of each parents DNA it is not exact (reference previous question Generational Loss in DNA testing) and will vary by sibling on what DNA segments are passed onto each child and it can vary.. I seem to recall easily up to a ~10% difference in shared cMs between two siblings in my very amateur research so far for some of my 1st cousins. This would account for the difference in your tests as well.
Where the chromosomes matching comes into play is that it can provide clues on what common ancestor you share.. which at close distances this does not matter so much but for more distant ancestors it for me is a clue of a potential common shared ancestor. Such as as I illustrated in the "Family Trait in a Shared Segment" question.. though I haven't found an exact reference as to what this is scientifically called.. I have learned that people who share a common ancestor sometimes retain a particular common trait unique to a particular chromosome. While far from bulletproof, I look for shared segments sometimes in a particular range in a chromosome to identify someone from that line. The biggest caveat with this method is I have to first establish a high confidence shared segment from multiple lines of a common known and documented ancestor from people who have tested.. but this helps me identify others from that line. I have done this with my primary research line who share a common couple born in 1815/1817.. and the 7 known and documented descendants (age 95 to 20) I have tested from 5 different lines of their children all have shared segments in a particular region of Chromosome 1. Beyond that shared segment they share no other segments over a length of whatever ftDNA's filter is (I think it is 5cms?).
On your third question, about longest matching cM.. yes each service uses a slightly different method for measuring what a match is as well as what a shared cM's minimum length is to be countered/aggregated on a match. Some services also do not count the X chromosomes in their calculation. Basically the important thing to know is that longer continuous cM segment is that the match is likely closer in nature because the segment is unbroken and been less diluted through the generations.