Specific Regions and Zones on GCP
I'm running an automated service that requires colocation with third parties to minimize latency. Currently, my compute is distributed across a couple of cloud providers (fly, GCP, AWS), but some components like a database and some analytics endpoints are hosted on railway.
Having multiple cloud providers is a PITA and is creating a lot of unwanted complexity and increased latency for certain operations. I wouldn't mind using GCP locations only, plus one AWS colo, because GCP locations should suffice.
Would it be possible to deploy my services within specific GCP regions and zones to decrease latency with third parties? What would that cost?
Thanks!
:sad_hamster:
@Brody
13 Replies
Project ID:
5ffdeb0b-59bc-4b91-9046-48f61f35249e5ffdeb0b-59bc-4b91-9046-48f61f35249e
what would be sufficient latency?
RTT <20ms for certain endpoints, but up to <300ms for slower ones
that should be achievable on metal, where are the locations you would be connecting to from our metal regions?
chicago, mazovia (poland), singapore, uk, hesse (germany), washington state
i think you could get pretty damn close to rtt 20ms for most of the locations
i think once we have our metal regions built out you give it a try
ok
:pray_joe:
:party_railway:
also @Brody how can i migrate db and cache to railway metal us east?
im gonna test some stuff on metal
metal does not yet support volumes
:sad_hamster:
soon