Firebase optimisation for highly requested queue

Question

A game based of firebase where every user push a request to play on the same waitingList so that a nodeJS server prepare a round game for every two waiting players separately. I am using Firebase JS v3 on nodeJS:

    var ref = firebase.database().ref();

//this is where we populate rounds
var gamesRef = ref.child('games');

//users path
var usersRef = ref.child('users');

//waiting queue
var waitingRef = ref.child('waitingList');

//listen for each new queued user
waitingRef.on('child_added', function(snap){ 
  waitingRef.transaction(function(waitingList){
    if(waitingList !== null){
      var keys = Object.keys(waitingList);
      //take two players
      if(keys.length >= 2){
        //prepare a new round for users
        var round = prepareNewRound(waitingList[keys[0]], waitingList[keys[1]]);

        //remove queued users from waitingList
        waitingList.child(keys[0]).remove();
        waitingList.child(keys[1]).remove();

        //create new game
        var roundKey = gamesRef.push(round).key;

        //notify users about the new game created
        users.child(waitingList[keys[0]]).child('current_round').set(roundKey);
        users.child(waitingList[keys[1]]).child('current_round').set(roundKey);
      }
    }
  });
});

The waiting list is highly requested by users that could exceed 10k simultaneous real time users. This way is performing faster than Firebase-queue but have repetitive slowness of 1 minute every 5 minutes where the waiting list become longer and longer before becoming normal later. Is there any optimisation that could be made to the concept?