When put in parallel, you split the current over multiple diodes. If the diodes are identical, then the current will be split evenly.

But the diodes are not linear like a resistor. If the diodes has different knee points - the voltage where they will light - then things will quickly become complicated. And the knee is very sharp. Below the knee point (Vf) it will take a couple of uA to climb the voltage a lot. Above Vf, every extra mV will represent a lot of extra current.

You may get one diode that will receive a significantly higher current. That LED will no longer have 30mA and be rated for 100,000 hours. It will (if the current imbalance is high enough) fail. If that diode breaks to infinite resistance, then you will pump the same total current through one diode less. Guess what - the other diodes will have more current than before. Now, if they also are a bit different, then one of them may get significantly more current and fail. See the bricks falling?

That is why you always have resistors when using diodes (unless you have a constant-current driver). This resistor - being linear - will take up the slack. If the Vf is just below 2V, then a little more voltage gets over the resistor. If the Vf is a bit above 2V, then a little less voltage over the resistor. But the current will not vary so much because of the resistance.

If you ave 5 diodes, each speced at 2V, then you can not use 10V to drive them. You need a higher voltage, since you still need a resistor as counter-balance. You may use 12-15V. The higher percent of the voltage you have over the resistor, the less affected the current will be with the variances in the diodes.

The advantage with the serial connection is that they are so easy to drive. Just one tiny little open-collector transistor for each set of diodes - no much current at all. And very little power losses. Spending 2V on the resistor and 10V on diodes means about 83% spent in the diodes. Having 2V diodes connected to +5V means 40% energy spent in the diodes.