Over the last year we have seen quite a lot of people that get worried about the UTXO growth. Lets find out why.
For those catching up; the UTXO is the Unspent Transaction Output database. The essential database that is needed to find out if a new incoming transaction is actually spending money that exists.
The growth is explained well in this blog from Gavin Andresen, the blog is already older, so the growth has continued.
There are some reasons why people claim this is an issue. Lets address each of them in turn.
The UTXO needs to be in memory completely. It will grow faster than internal memory can.
Answer: The implementation in all Bitcoin full nodes is that the entire database is stored on disk. With a smart memory-cache to speed up lookups. The memory-cache is configurable and the default is 300MB. (it's the
dbcache config option) The entire database is significantly larger than 300MB, as the above blog states.
The claim that it has to be in memory is weird because none of the full node implementations actually do that. Sure, miners can adjust their settings to allow a bigger cache, and be different from everyone else. But this doesn't mean the UTXO should stop growing.
Bigger blocks make the UTXO grow faster!
Answer: The UTXO is a database of where money is stored. With more people using and holding Bitcoin, this will grow. The UTXO will not just grow based on more transactions, the UTXO will grow because we get more customers. Which can only be done when we have bigger blocks.
So the argument isn't directly false, it is misleading. It is missing the fact that UTXO growth is based on more people using Bitcoin. Maybe the people making this accusation are aware that saying they do not want more customers will make them look stupid.
UTXO on disk will make miners lose money because of slow block validation.
Answer: This may have been true in the past, but is no longer true since we introduced xthin blocks. This technology allows any node to communicate a new block with only the header and some meta-data. The receiving node will then use the transactions that are in his memory pool to reconstruct the block.
The direct result of this is that when a new block is mined miners will not validate the vast majority of transactions again because the transactions in their mempool were already validated. This means that there is no bottleneck any more for miners using xthin (available in BU and Classic).
Bottom line is; it's a database. MySql/PostgreSQL/MongoDB etc. have had decades to perfect database technology. Being scared of a database growing on today's hardware is unfounded.