to find a pool. should have at least one backup, probably two. lower latency isn't always best, check what your DOA rate would be & check the pool's orphan rate. a pool that gets 10% orphans that you have 20ms latency to would probably be lower efficiency for you then a pool that gets 5% orphans w/ 150ms latency (this likely wasn't the case when the shares were 15s apart instead of 30s, but it reduces the importance of low latency)
a few words about latency. there is a perception that 100-200ms latency is batter than 20ms. as it is explained in
efficiency thread that because of limiting number of included tx and as well - lower tx fee.
That is mistake! 20ms (lower) IS better then 100ms (higher) latency.
Let's look. It's true that to reduce latency is used limiting of tx number
For example we can include only transaction that have 0.0001 and higher tx fee
this way we will filter 'transaction dust'
what we loose in this case?
avarege number ot transaction in block now is about 300-500, so filtering dust we reject about 0.0001*500=0.05Bbtc fee
and it will be only if all 500 transactions have less than 0.0001 fee. Practically it hundreds times less.
So rejecting about 0.0005 fee each block we get latency 20ms instead of 200ms and theoretical less amount of rejected shares on pool
obviously lower latency is better than higher latency... as in network latency.
bitcoind getblocktemplate latency would cause you to have a higher chance to have an orphan BLOCK, but not really a whole lot else (though generally when you have a high gbt latency as in 0.5s+, your whole damn system is strained). already, 1/2 the time or more, these shares that are being 'punished' end up being the ones that become the main chain.
the problem with having a lot of transactions on a public node that a lot of people use, is that with 5-10 work requests, it can take 100ms to get these all out if you have a 100kb block, even on something like a xeon e3-1270v2. if i was running a private node w/ just myself, i would set maxblocksize to 500000
i said if you choose a p2pool, the # of orphans that p2pool has is the most important thing to look at. the DOA rate is easy to see, run a miner on it for 10 minutes
"82.196.8.44:9333": 3816
so you are running 3816 bytes of transactions atm. i am at 4873 (though I think I'll just go ahead and drop that to 2000 or so)
my latency is average of 3.16ms over the last 24 hours, yours is 7.64ms, so you don't need to lecture me about it. i do this so there is less time between work assignments, not for getblocktemplate stats