If you are using streaming replication, then you might come up to some numbers with a calculation like:
number of blocks changed over time * block size
block size is by default 8 KB
That will tell you how many KB you need over the specified period of time. On top of that you might want to also add the vacuum traffic, if you have scheduled one over night/weekend/rest time. If you are doing a daily vacuum, then probably you have to take into account that too, which probably goes way faster then the application write ratio. (but you might not care if over night your replica is not in sync..)
When you come up with a number, consider it as the bare minimum, therefore a good size for it might be 2x or 10x depending on your growth ratio over time and the confidence you have in your calculations.
The topic as you see is not easy to describe in a few words, I suggest you to be pragmatic and come up with a load test. Load test is usually a great idea also to see if your machine is properly sized for your production workload.
I hope it helps.
regards,
fabio pardi
On 10/21/18 3:01 AM, prasanna rk wrote: > Can you please share some document link for referring the prerequisite > of network configurations. > > On Fri, 19 Oct 2018, 18:41 prasanna rk, <prasanna2mail@gmail.com > <mailto:prasanna2mail@gmail.com>> wrote: > > Thank you Fabio.. > > Can you please let me guide me with the link or document that I can > refer for this.. > > On Fri, 19 Oct 2018, 17:05 Fabio Pardi, <f.pardi@portavita.eu > <mailto:f.pardi@portavita.eu>> wrote: > > Hi Prasanna, > > that depends very much from how much data you write on master. > > If you write more, then you will need more bandwidth. You also > might have peaks in writes, but also be keen to wait some time > for standby to catch up with master. > > Also worth noticing is that streaming replication and logical > replication have different network usage requirements. > > regards, > > fabio pardi > > > On 19/10/2018 12:22, prasanna rk wrote: >> Hi Team >> >> What should be the minimal network speed between source and >> target to setup replication. >> >> Regards >> Prasanna >