The size shouldn't differ between sorted and unsorted files.

My bam file is 26.9 gb

Maybe this is sam file? and what you're refering as sorted bam is bam file. But, 30Gb sam should be compressed to a smaller file than 18Gb.

I suggest you post the commands you used as well as the what you're sequencing.

It can be slow, but I don't have an exact time. It's easy for you to get this though and it'd be more accurate than anything people here can give you as hardware differs. Index it and time it on a smaller region. Extrapolate from there.

A few pointers though.

• Don't use samtools mpileup to make bcfs. It's ancient. Use bcftools mpileup instead.
• Use a modern bcftools, particularly if you have long-read technologies. (There have been substantial changes in develop branch too for accuracy.) Also specify the instrument profile with -X. This can dramatically improve indel calling recall and precision.
• Mpileup is an embarassingly parallel problem. There isn't a built in wrapper than I know of, but it's not hard to do yourself. Just produce a list of regions (could be 1 per chromosome, but better to do smaller fragments) and run bcftools with gnu parallel to process each region and spit out a new file. You can then just bcftools concat these together in order to get the final mpileup.
• Also see http://www.htslib.org/workflow/wgs-call.html, although I think some of that is rather legacy (eg I'm not sure if IndelRealigner still exists, or whether it's still even necessary any more).