Aerospike Query Return Highest Value

Question

I'm trying to create a query for my Aerospike database, that would return the highest value in a specific bin; similar to the way that the MAX() function works in MySQL. For example, if I had a set like this:

+--------------+---------+
| filename     | version |
+--------------+---------+
| alphabet.doc | 4       |
| people.doc   | 2       |
| alphabet.doc | 6       |
| people.doc   | 3       |
+--------------+---------+

What I need is to only return the filename with the highest version number. At the moment I can add a filter like this:

    stmt := db.NewStatement(DBns, DBset, "filename", "version")
    stmt.Addfilter(db.NewEqualFilter("filename", "alphabet.doc"))

    // run database query
    records := runQuery(stmt)

Anyone know how to do this?

Answer 1

You can apply a Lua user-defined function (UDF) to the query to filter the results efficiently.

E.g. here is a Stream UDF that would return the record with the max. version number:

function maxVersion(stream, bin)
  -- The stream function cannot return record objects directly,
  -- so we have to map to a Map data type first.
  local function toArray(rec)
    local result = map()
    result['filename'] = rec['filename']
    result['version'] = rec['version']
    return result
  end
  local function findMax(a, b)
    if a.version > b.version then
      return a
    else
      return b
    end
  end
  return stream : map(toArray) : reduce(findMax)
end

Using the Go client you would execute the function like this:

  stmt := NewStatement(ns, set)
  recordset, _ := client.QueryAggregate(nil, stmt, "udfFilter", "maxVersion")

  for rec := range recordset.Results() {
    res := rec.Record.Bins["SUCCESS"].(map[interface{}]interface{})
    fmt.Printf("filename with max. version: %s (ver. %d)\n", res["filename"], res["version"])
}

I've uploaded a fully working example as a Gist here: https://gist.github.com/jhecking/b98783bea7564d610ea291b5ac47808c

You can find more information about how to work with Stream UDFs for query aggregation here: http://www.aerospike.com/docs/guide/aggregation.html