Replies: 2 comments
-
I have built a docker image for ib-gateway and TWS. this gets us closer to "one-click deployments" I packaged a research environment too, it's a jupyter docker image with commonly used packages for quant trading. I spend most of my time here. |
Beta Was this translation helpful? Give feedback.
-
Experimenting with the ibkr gateway on Linux with a minimal viable setup where it can restart itself and we can still login to it (no IBC, and maybe one day not even the java gateway). Current working attempt:
No need for heavy gnome or kde, just a simple self-contained window manager running a single application which persists in the background with TigerVNC remote access (accessed over ssh port forwarding). Of course the remaining problem is the gateway still only survives the daily restart and can't login by itself again — but, I think it would be easy to write a vnc client to read the vnc session every second and, if the vnc session is showing the gateway login screen, enter username/password and click the login button just using python automation. More progress soon. Also potentially looking at evicting the gateway completely? The gateway has "encryption optional" mode, so it seems we could just capture the traffic directly and forward things ourselves in a custom non-java, non-GUI proxy application too? Could be interesting to try. |
Beta Was this translation helpful? Give feedback.
-
AWS has a "Local Zone" in NYC they refreshed in January 2025 (
us-east-1-nyc-2a
).I gave it a try and I'm getting a 4 ms ± 100 µs ping to the IBKR gateway address ndc1.ibllc.com which is better than anywhere else I've tried.
My current plan is to redeploy an initial architecture running the standalone IBKR gateway in
us-east-1-nyc-2a
on onem6i.large
instance (around $90 per month), with trade execution code also on that instance, but with analytics code running inus-east-1
on cheaper cost-performance-tradeoff spot instances (the NYC local zone has no spot instances currently).Why outline deployment details? For accessing markets, there's optimal locations and parameters for being close to the market execution computers at an affordable cost while also still being able to manage your environments and executions easily.
We should eventually be able to turn this into a "one-click infrastructure deployment" strategy for people who have enough technical capability to run AWS setup automation on their own.
Anybody else have other interesting deployment strategies or infrastructure architectures out there?
Beta Was this translation helpful? Give feedback.
All reactions