I guess you are asking how to implement it in EXCEL. Here is one way:
Type         Year        Sale        Cum_Sale
A        1        5        =SUMIF($A$1:A2,A2,$C$1:C2)
B        2        6        =SUMIF($A$1:A3,A3,$C$1:C3)
A        2        7        =SUMIF($A$1:A4,A4,$C$1:C4)
B        1        3        =SUMIF($A$1:A5,A5,$C$1:C5)

It seems that some Year 2 data are ahead of Year 1. Are you allowd to sort the data based on Year first? If not, it seems hard to use sumif here. However, it is easy to implement it using VBA to write a few lines of code.

My bad. You can do it using sumifs as follows:
Type         Year        Sale        Cum_Sale
A        1        5        =SUMIFS($C$2:$C$5,$A$2:$A$5,A2,$B$2:$B$5, "<="&B2)
B        2        6        =SUMIFS($C$2:$C$5,$A$2:$A$5,A3,$B$2:$B$5, "<="&B3)
A        2        7        =SUMIFS($C$2:$C$5,$A$2:$A$5,A4,$B$2:$B$5, "<="&B4)
B        1        3        =SUMIFS($C$2:$C$5,$A$2:$A$5,A5,$B$2:$B$5, "<="&B5)

哥，我用的是SAS，不是Excel。而且我的表是很大的，没可能一个个写吧

data have;
input type $  year  sale;
cards;
A        1        5
B         2        6
A         2         7
B        1          3
;

data wanted;
  if 0 then set have;
   *** define Hash object;
   declare hash h(dataset:'have' ,ordered: 'Y');
   declare hiter hi('h');
        h.definekey('type','year');
                h.definedata('type','year','sale');
                h.definedone();

   *** calculate cumulative sale;
        rc=hi.first();
           _type=type; _sale=sale;
       do while(rc=0);
              rc=hi.next();
                  if type=_type and rc=0 then do; sale=sum(sale,_sale);  h.replace(); end;
                  _type=type; _sale=sale;
                end;

   *** add the cum_sale to original dataset;
   do until(done);
      set have(rename=(sale=orig_sale)) end=done;
          h.find();
          output;
          rename orig_sale=sale sale=cum_sale;
          drop _: rc;
    end;
run;