opolis / exchange Goto Github PK

Beyond the work to be done in #4, this service would ideally connect to a fee provider to dynamically adjust fees as needed based on time preference and network congestion. It would probably be wise to incorporate some configurable upper limit too so it doesn't run out of control in times of very high fees.

Right now, we are stuck at 10 gwei for our ETH tx fee outbound to Wyre. This doesn't work well at all in a dynamic fee marketplace. It should be quick and easy to change network fees.

Or rather, make it optional. Ideally, users have the option of running a self-hosted auxiliary service (to be built) that will send this service webhook events for tx confirmations using data from their own node. Not only is this more secure (webhook service can authenticate via HMAC digests), it's less risky as we won't be dependent upon BlockCypher to stay up and running.

Right now, this service utilizes manually configured "payers" to associate new addresses with, all under a single key mnemonic with varying HD paths to determine the actual address to sign for.

Ideally, this service provides an authenticated API for registering new payers (an action they could take on their own via a dashboard), and generating new, unique keys for each payer, while allowing them to generate one-time addresses for payment.

RDS is not only expensive at scale, but it currently requires a fair amount of manual work to configure it and link it into this service deployment. I think the overall schema is simple enough to where something like DynamoDB would work.

Dynamo provides a couple key benefits

1. can easily be configured and provisioned quickly via CloudFormation
2. authentication can be done entirely by IAM policies, meaning no passwords have to be configured

Wyre supports BTC, ETH, and DAI as incoming crypto. DAI support in this service would be a nice improvement.