Black is in fact the first to get into real trouble, with the premature ...c5 pawn break. This is a repeated conceptual error of mine (as in Annotated Game #62) and a major learning point from the game. Black's subsequent lack of development and poor protection for his king in the center gave White a real opportunity to put more pressure on. However, by move 15, Black manages to fully equalize and passes the danger zone. Now White decides to play too optimistically for a win, disdaining an initial queen trade and then finally being forced into one under less favorable circumstances. The next turning point comes when Black pressures the isolated d-pawn and White fails to protect it adequately due to a tactical pawn break. An interesting point of technique by the move 24 variation, in which White voluntarily gives up the d-pawn in order to shatter Black's pawn structure and achieve a level ending.
Despite Black's winning the d-pawn, he soon fritters away his advantage, being overly concerned about White's rook play on the g-file. After rooks are exchanged off into a drawish knight and pawn ending, White for some reason essentially deactivates his own pieces, allowing Black to centralize his king and obtain passed d- and a-pawns, giving him a won game...if only Black had pushed his passed pawns. Black fails to advance one to gain a crucial tempo, then White forces the draw.
Aside from the lesson of the premature ...c5 break, my main takeaway from this game is the value of piece activity in the endgame and some practical experience in analyzing N+P endgames. The opening transposition is also worth some consideration.