Get a histogram of the count of unique values in elasticsearch

I have docs that looks like this in ES:

{
  "ipAddress": "w",
  ...
},
{
  "ipAddress": "x",
  ...
},
{
  "ipAddress": "x",
  ...
},
{
  "ipAddress": "x",
  ...
},
{
  "ipAddress": "y",
  ...
},
{
  "ipAddress": "y",
  ...
},
{
  "ipAddress": "z",
  ...
},
...

I'm looking to understand the distribution of the number of events per ip address. In other words, for the data above, i want the results:

numOfEvents | numOfOccurences |
_______________________________
     1      |       2      (2 times, w and z)
_______________________________
     2      |       1      (1 time, y)
_______________________________
     3      |       1      (1 time, x)
_______________________________

The cardinality of ipAddress is very large, so it's infeasible to get a terms aggregation by ip and build the histogram on the client side. I saw this question which seems to indicate that this isn't possible? A transform table is the only option? We're on an older version of ES that we need to upgrade to use this, and I would imagine using this transform approach would use up a lot of space (currently about 2TB of data).