Now that with v5.6 and above, since ES is moving towards removing the concept of mapping type and essentially with one index holding only one type (unless we define our own type attribute), does it really make sense for someone with a smaller data volume requirement to have 5 shards per index?
For example, we are using ES to perform full text search on RDBMS data. So we will have around, let's say, a million docs coming to 1 GB index size. Now does it really make sense to use the default 5 shard per index setting, if my data volume growth is very limited (say 5-10% per year)? Trying to understand what is that I would gain by having 5 shards for such a small index size.