How does the bandwidth delay product optimization apply to long-distance SnapMirror replication in ONTAP?
Answered at 14:18 in Episode 414 - ONTAP Networking Updates - Fall 2026 · TechONTAPPodcast
How does the bandwidth delay product optimization apply to long-distance SnapMirror replication in ONTAP?
For long-distance asynchronous SnapMirror deployments, TCP must be tuned with a receive window scaled to the bandwidth times latency. For example, with 100 milliseconds latency and a 10 gigabit link (1.2 gigabytes per second), the receive window requires 125 megabytes to fully utilize the link. This large pre-allocated memory for TCP sessions can degrade system performance if applied globally, which is why dedicating an IP space for long-distance replication—as recommended in SnapMirror best practices—allows independent optimization without impacting local client access or buffer cache.
Full answer and detailsRead the whole answer, key facts and every other question from this episode
In short
For long-distance asynchronous SnapMirror deployments, TCP must be tuned with a receive window scaled to the bandwidth times latency.
Richard Schefferniger, Solution Architect, EMEA and Latin America Region at NetApp · Episode 414 - ONTAP Networking Updates - Fall 2026 · TechONTAPPodcast · answer at 14:18
Key takeaways
- For long-distance asynchronous SnapMirror deployments, TCP must be tuned with a receive window scaled to the bandwidth times latency.
Quote to remember
“IP spaces or VRFs is the equivalent of your three yard fence, right? So it's much more hard to step over a three yard fence than it is over a one yard fence.”
— Richard Schefferniger
The full answer · about 2 min
For long-distance asynchronous SnapMirror deployments, TCP must be tuned with a receive window scaled to the bandwidth times latency. For example, with 100 milliseconds latency and a 10 gigabit link (1.2 gigabytes per second), the receive window requires 125 megabytes to fully utilize the link. This large pre-allocated memory for TCP sessions can degrade system performance if applied globally, which is why dedicating an IP space for long-distance replication—as recommended in SnapMirror best practices—allows independent optimization without impacting local client access or buffer cache.
Key facts
- VLANs introduced in networking technologies approximately 35 to 40 years ago
- IP spaces have been available in ONTAP since approximately version 8.0 to 9.0 releases
- CFM broadcast detection checks occur approximately every 10 minutes per port
- CFM broadcasts must fail twice in a row (approximately 30 minutes) before ONTAP detects broadcast domain fragmentation
- New TCP stack in ONTAP 9.19 delivers 3 to 4 times performance improvement for remote workers, sometimes even higher
People, companies and topics mentioned
More answers from this episode
- 3:00What are the main use cases for VLANs in ONTAP networking environments?
- 8:42How do IP spaces enable secure multi-tenancy in ONTAP systems?
- 18:18What TCP optimizations does ONTAP 9.19 provide for remote workers accessing SMB home shares?
- 22:48How do failover groups work within broadcast domains to ensure performance during link failures?
- 23:06What is the CFM protocol and how does it improve broadcast domain detection in ONTAP?