Skip to main content

NetApp_Insight_2020.png 

NetApp Knowledgebase

What are the implications of setting DNS TTL with FabricPool?

Views:
100
Visibility:
Public
Votes:
0
Category:
storagegrid-webscale
Specialty:
sgrid
Last Updated:

 

Applies to

StorageGRID 11.x.x and later
Data ONTAP 9.2 and later
All AFF Platforms
FabricPool

Answer

Affects of TTL with DNS RR Group

  •     TTL weights the amount of time before the next DNS record is resolved
  •     For Instance:
    •         TTL value of 4 hours
    •         Host1 handles all IO for 4 hours straight
    •         After 4 hours Host2 handles all IO for the next 4 hours

DNS TTL (Time to Live)

  •     Applies to each DNS record
  •     Dictates how long the DNS query is cached for the specified record 
  •     Speeds up name resolution
  •     Default Value: 12 hours or 24 hours
  •     Value of 0 does not cache the query

DNS RR (Round Robin) Group

  •     Consists of multiple DNS records
  •     Alternates resolution of records one at a time

DNS RR in FabricPool with StorageGRID

  •     DNS RR is used to create High Availability Gateway Nodes
  •     Gateway nodes establish HTTP sessions to send S3 operations to the Storage Nodes
  •     TTL values can cause a single Gateway Node to establish all sessions to serve IO
    •     Performance Penalties can be seen due to Single Host serving all IO

Additional Information

Choosing TTL values

StorageGRID HA Groups

  • Introduced in StorageGRID 11.3.0 release
  • Recommend HA groups starting with StorageGRID 11.3.0
  • Uses Virtual IPs (VIPs) to associate to the Active Node
  • Does not require DNS RR for HA
  • Failover Monitoring triggers failover for the following reasons:
    •         Active Node on which the interface is configured goes down
    •         Active interface goes down
    •         Load Balancer service stops on Active Node
    •         HA service stops on Active Node