topic Re: Async Replication - Performance Expectations in ONTAP Discussions

Async Replication - Performance Expectations

sandeepthots — Wed, 04 Jun 2025 10:14:24 GMT

Need a clarification....We have proposed AFF A400 with 450TB effective capacity and will be using Async replication to DR.
Distance is 1200Kms and Bandwidth is 100Mbps

Proposed Connectivity is 16G FC

How much RPO we can expect from asynchronous replication as per best practices

Re: Async Replication - Performance Expectations

paul_stejskal — Thu, 09 Sep 2021 12:26:12 GMT

If your network is healthy it should peg it 100%. Usually replication can see 500-1000 MB/s (Bytes, not bits) with enough horsepower (A400 counts) and concurrent SnapMirrors.

Re: Async Replication - Performance Expectations

shamz — Mon, 20 Sep 2021 20:41:25 GMT

So you want to have a DR replica of 450TB over a 100 "Megabit per second" link?

I have no idea what "Proposed Connectivity is 16FC" has to do with things here.

Basically, you will need to get a truck and move the DR system to the Production site to ever come online again in case of a total disaster because the bandwidth (if I understand 100 Megabits per second correctly) is insuffucient to restore that data in a reasonable amount of time.

An ideal example with 100Mbit/s would get you 45GB per hour, so you would need 416 days of perfect full power bandwidth to restore 450TB. Now there are things like compression available, but how well that will work is sort of a matter of what kind of data you are storing. 1200km adds a bit of variance to the equation as well, but mostly if the quality of the link is poor. I think snapmirror and ONTap can adjust the TCP send and receive windows sufficiently.

You have a bit of a math assignment to do.