Why is a scrub of a pool with 98 4TB drives taking more than 6 days long?

I'm doing a scrub of a pool on a ZS3-4 that contains 98 4TB drives.  The scrub was started on January 19th.  Here's what zpool status -v says about the pool this morning (1/25):

  pool: RAIDZ2-pool-a

state: ONLINE

  scan: scrub in progress since Fri Jan 19 13:21:29 2018

    8.05T scanned out of 156T at 16.3M/s, (scan is slow, no estimated time)

    0 repaired, 5.16% done

The ZS3-4 is used for Exadata backup.  It has not been heavily stressed over the past week, as the analytics below show:

What else should I be looking at as causes for a scrub running this slow?