Update
As I (SteemWorld) cannot survive for much longer with the given STEEM price, I need to take action now. So, I started a steemd instance (with MIRA enabled) on my dedicated machine and will enable my witness in the coming days. I don't want to be an auto-witness out of a pre-packaged box, so I decided to build all things that I need myself.
I know that MIRA might not (yet) be the way to go for a witness, because replay times take a bit over 3 days (in my case), but I want to use it anyway. In any case, I want to use MIRA in the future for the SteemWorld back-end, so I think it's a good idea to work with and experience its behaviour now already. Learning by doing is always the best way for me.
If I should miss blocks or experience any difficulties while running it on that (just slightly busy) machine, I will think about other solutions. Fact is, I cannot afford an additional server, so it has to work.
The good thing is, that I now can receive blocks and other data that I need for the SteemWorld databases directly from localhost and don't need to rely on other nodes anymore. In my tests it worked perfectly (so fast!) so far and I begin to love this thing called 'MIRA' :)
As I saw recently, Steemit has been working on a Rewards API/Plugin for steemd and maybe I can use this in future to improve the loading times for the coming rewards. Being able to see the coming curation rewards with a click in a few seconds? Exciting times ahead!
Maybe I can even build my own plugins for steemd, so that the front-end devs out there don't need to reinvent the wheel again and again... Maybe I can implement my planned SDS (at least partially) directly inside the node, so that we could use a compressed but at the same time fast readable and writable block_log...
I'm broke, but at the same time I am very fascinated about the new possibilities :)
More news to come soon.
Steem On!
