All of the information that steemreports tools use comes directly from the blockchain's RPC servers. These are expensive machines to run, mainly because of their large RAM requirements.
This has concerned me as a possible scalability issue for a while because the RAM requirement has also been growing very quickly. So I was pleased to discover today when looking at the steem code (as I do every so often), that there is a change currently being made which I think should significantly reduce the RAM required to host a full API node, and also how much it increases over time.
The largest use of RAM in an RPC node is the account history index which is used for many important things such as showing your lists of: blog posts, comments, replies and transactions.
This index currently allows steemit.com (or any other front-end) to show a single account's entries, going back to the start of the blockchain, but the change allows node operators to limit this to just the last 30 days if they wish to save RAM.
I believe the RAM required to run a full API node should be reduced from around 128GB to less than 64GB, and will then grow at a fraction of it's previous rate.
I'm not quite sure how the older data will remain accessible for those who need it, but it may be provided through the Jussi reverse proxy and Hivemind database projects, which are currently being worked on.
This change will not come into effect immediately, but I am pleased it is being addressed as it makes running these servers more affordable, and keeps the network more decentralised.
