You are viewing a single comment's thread from:

RE: SMTs Release Dates!

in #steem6 years ago (edited)

No real surprise here... A announcement for SMTs which won't happen until 2019 (who knows when), meanwhile there are still serious scaling issues that need to be solved for the steem platform. Why not devote time to making it so that full nodes do not require almost 1TB of RAM to run? That doesn't seem to be a priority despite the fact that it would make the technology miles and miles more desirable then SMTs since the features all require better scalability. Steem desperately needs sharding or some kind of asynchronous system.

Sort:  

I don't think that fixing a issue that only affects witnesses and can be easily fixed adding RAM should slow down the development of a feature that will allow a huge number of new use cases

It doesn't only affect witnesses. Any application that runs on top of the blockchain is affected by this issue. The only way to access the RPC of the steem blockchain is to use a Full Node. Any performance issues that come from this lack of scaling proliferate down to all of the users. It makes developing for the platform harder as well when the API and RPC are not stable. If they implement SMTs on this platform as it is it will cause tons of issues for all of the end users because the platform doesn't scale.

Just consider these numbers for a moment... According to the latest steemit statistics post the platform hosts between 50 and 60k concurrent active users and a full node on the blockchain requires almost 1 TB of RAM to run smoothly. Bitcoin, the largest blockchain in the world is currently supporting 25 million wallets (and for the sake of argument, let us just say that 5 million of them are active users on a daily basis). A full Bitcoin node requires only 2 gigs of ram to run properly. Bitcoin cash which also has its own blog type platform, memo.cash, also only requires roughly 2-4 gigs of ram to run a full node. While admittedly none of these platforms are identical with one another in how they operate, the cost of performance vs the amount of users supported is absurd.