It isn't that difficult, really. I mean, there is implementations of ethereum full nodes in like 10 different languages. There is just a need to reimplement.
The reimplementation, at minimum, needs to split the user registry, the token ledger and the forum/voting system, and the underlying witness/cluster membership management system need to run separate database instances, using a more suitable database.
Note that this will only allow Steem to maybe cope with up to about 1 million active users (currently around 15,000). To cope with a rapid influx of new users beyond this level, requires some massive database engineering feats the likes of which to date, Google and Facebook are the only organisations that have pulled this off.
As @andybets was telling me, he has most of the chain's posts and comments extracted into a flat file database, which is still better than it being embedded within a graph database.
There is the resources and knowledge in the community to make the necessary changes, and I expect these changes will be forthcoming over the next few months.
I am just hoping though, that someone will do something that I think should be done. Erase the premined SP, and release the vests to the rewards pool. The remainder of accounts are invested or earned stake. The profile of trending posts will entirely change, it certainly won't have Dan Larimer's opining at top trending because he can vote so much on himself that even half the other whales can't flag it down, if they work together...
Thank you for your honest opinion, it means a lot to us.
Doesn't hurt me at all. My node doesn't need near as much memory as a service that needs to track account balances like, you know... an exchange.
If you want to see Steem stick around and even potentially make money with it, what you're doing is so short sighted it's like a dog eating its own leg because it's hungry.