Hello Steemians, welcome to our latest Steemit Engineering Update. You can view our last engineering update here.
We are continuing to let the wallet application soak so that people can become familiar with steemitwallet.com. During this process we are continuing to test and improve the Social Condenser application we have running in our development environment. We have also rolled out a number of small improvements to Condenser. We merged in the private key backup pdf we mentioned in our last update which will help ensure that people have a copy of all their private keys. Many users are solely using their Master Passwords when using steemit.com and only have that password stored in-browser. This exposes their Master Password to a malicious actor and also creates a risk that they will lose access to Steem if they lose access to their computer for whatever reason.
On Monday we released the sidebar version of our “Coin Marketplace” so that people can watch their favorite cryptocurrencies without leaving steemit.com. We plan to improve this feature over time, including through the addition of a larger marketplace which will feature the top 200 coins. We will also look to migrate the backend service provider to 
We have now run a number of MIRA tests in our production environment! We are following a rollout similar to the approach we took with Hivemind which led to an extremely smooth transition. That process involved running the software in our production environment temporarily, accumulating data, and then rolling back the changes regardless of the outcome. We then analyze that data, determine what, if any, changes need to be made, rinse and repeat.
While we are running these live tests, we are continuing work on our hybrid-in-memory reindexing. RocksDB is a great database for enabling us to get the kind of performance Steem apps require but on commodity-hardware. However, those benefits only accrue after reindexing; the rebuilding of the database from scratch. For that function, RocksDB is not optimal.
There are certain parts of the reindexing process that are so resource intensive that no level of optimization can enable commodity hardware to rival the speed of something like an NVMe. By having MIRA perform those specific processes in-memory only during reindex, we believe we can approach our previous reindexing times with a significant cost reduction. Most importantly, once reindexing is complete the additional RAM demands will terminate. This will be hugely valuable for anyone looking to perform a reindex on commodity-hardware.
It is extremely satisfying to see these important projects nearing their conclusion. These projects were necessary in order to ensure our survival as an organization. The speed at which we have been able to execute on the tasks and the degree to which we have been able to reduce our run costs, have exceeded our expectations. This, combined with the ad revenue we have begun generating, makes us very optimistic about our future as an organization.
While we were in “survival mode” we now feel confident in our sustainability going forward and are eager to begin thriving. We look forward to sharing our strategy for this new phase of growth as it is developed. We want to thank everyone for sticking with us during this difficult time, but we now feel that the hardest parts are largely behind us.
Full Steem Ahead!
The Steemit Team
