00:13 <stellar-slack> <lab> i want to run a node

01:23 <stellar-slack> <arctaurus> I'd like to know the Hw-Sw requirements to run a node. Any link with this info?

01:36 <stellar-slack> <lab> every 100k account + 100k transaction will cost 1GB storage, my est.

01:36 <stellar-slack> <lab> the requirements depends the network volume.

01:43 <stellar-slack> <lab> history archive is about 1/3~1/4 of db storage

01:43 <stellar-slack> <irisli> many of the the 100k transactions will be in the past (stored in history archives) and won't be part of the current ledger

01:44 <stellar-slack> <lab> is it safe to valcumm the txhistory table?

01:46 <stellar-slack> <lab> IMHO current ledger size just reduced memory requirement.

01:46 <stellar-slack> <jed> yeah you don't need the past

01:49 <stellar-slack> <lab> i saw average cpu/mem (several GB) can sustain a node

01:52 <stellar-slack> <lab> so c4.xlarge and above is suitable to run a node

01:52 <stellar-slack> <lab> i prefer c4 because it has higher per core ecu

01:53 <stellar-slack> <lab> stellar-core is single thread

02:00 <stellar-slack> <lab> regarding sw, there is no external requirement except postgres if you choose it.

02:01 <stellar-slack> <lab> and maybe tools to upload history archive to cloud storage.

02:02 <stellar-slack> <lab> I think the top challenge to maintain a node is security.

02:03 <stellar-slack> <lab> the validation_seed is stored in plain text in your server.

02:05 <stellar-slack> <lab> matschaffer, can you share us more experience?

02:18 <stellar-slack> <matschaffer> lab: still early, but I'm leaning toward m3.xlarge. The extra cpu on c4 class is nice, but it comes at the expense of ephemeral volumes.

02:18 <stellar-slack> <matschaffer> EBS only doesn't seem like a great move for something with this much disk activity

02:42 <stellar-slack> <lab> I got your points, thanks.