Ah, thanks for these links everyone, very interesting and this clarifies a few things in my mind. Looking at this wiki page:
https://en.bitcoin.it/wiki/Transaction_fees#Technical_infopriority = sum(input_value_in_base_units * input_age)/size_in_bytes
So this is my transaction:
https://blockchain.info/address/18AA4H7Vm84NwqhxSE6HBpY6663b7uycoGSo as of the time of writing its queue value is approximately: 13,573,000 * 120 / 500 = 3,257,520
Assuming 6 blocks per hour and that my transaction has been queued for 20 hours and a size of 500 bytes as it is coming from one address. Is this correct?
I'm also assuming here that 'confirmation' in this article means the number of blocks that have passed since that transaction was broadcast to the network?
...for transactions which draw coins from many bitcoin addresses and therefore have a large data size, a small transaction fee is usually expected.
I understand that exchanges like MTGOX and Vircurex don't have separate wallets per users but effectively have their own wallet per exchange and then track a user's coin allocation via some form of database. So my question is do exchanges or even wallets in general make a point of issuing Bitcoins from the lowest possible number of addresses at all times when initiating a transfer to keep block size as low as possible at all times?
Is there a development roadmap in future to split up the blockchain storage so that no single node needs to store the entire data set but a set of nodes can have the data reference in such a way that a full blockchain can always be created from a set of n nodes ala RAID 5 where say 5 HDDs can have enough redundancy to loose a disk whilst still providing the full data set. This would allow larger block size without filling everyone's HDDs with a full copy of the blockchain per node.