clearvars

n=[8 9 9 9 9 9 9 9 6 8 9];

pval=1;

for i=2:11

  pval(i)=mod(pval(i-1)*10,101);

end

dec=1;

for i=2:11

  dec(i)=dec(i-1)*10;

end

totVal=mod(sum(n.*pval),101);

sum(n.*dec)

found=false;

while ~found

  for i=4:9

     if n(i)<n(i-1)

       n(i)=n(i)+1; n(i-1)=n(i-1)-1;

       totVal=mod(totVal+pval(i)-pval(i-1),101);

       if totVal==0

         found=true;

         break

       end

     end

  end

end

disp(sum(n.*dec)) 

fails to find an 11-digit solution. (The procedure was terminated manually as only finding an answer would terminate it by itself, and found to have gotten to the highest value.)

clearvars,clc

n=[8 9 9 9 9 9 9 9 6 0 8 9];

pval=1;

for i=2:12

  pval(i)=mod(pval(i-1)*10,101);

end

dec=1;

for i=2:12

  dec(i)=dec(i-1)*10;

end

totVal=mod(sum(n.*pval),101);

sum(n.*dec)

found=false;

while ~found

  for i=4:10

     if n(i)<n(i-1)

       n(i)=n(i)+1; n(i-1)=n(i-1)-1;

       totVal=mod(totVal+pval(i)-pval(i-1),101);

       if totVal==0

         found=true;

         break

       end

     end

  end

end

disp(sum(n.*dec))

does find a 12-digit solution:

The program stores the lowest number with the appropriate beginning, ending and sod. It's stored "backwards", units position first. The positions after the terminal 89 (backward remember) are examined until an decrease in the digit value is found compared to the previous. The previous is decremented and the current one incremented, maintaining the sod. The value of the whole n is adjusted using the modular values of those positions. Once a value mod 101 of zero is found the algorithm stops.