0

We have an exchange 2013 DAG with 2 sites and 4 servers. Each server contains 4 databases. One database is active on each server and 3 are passive (each server contains a copy of all 4 databases)

Periodically (maybe 2-3 times per week) there will be an unexpected network spike where two servers (between sites) will exchange data at our site-to-site link speed (50mbps) for about 15 minutes. During this time there is severe performance degradation (in outlook for users on those servers and on the network due to the link saturation)

My question is two-fold:

  1. What causes this large exchange of data between the servers?
  2. Is there a way to prevent it from happening or cause it to not saturate the link without implementing QoS on the link?

Thanks!

Abraxas
  • 1,229
  • 1
  • 16
  • 25
  • Why didn´t you simply use the Exchange Cache mode in the Outlook Client? Then the users shouldn´t see the "performance" issue as everything is cached locally or are they using OWA? – BastianW Feb 14 '17 at 13:36

1 Answers1

0

There is nothing that I can think of that would cause that to happen regularly. Therefore you need to check the database replication status when it is happening. It could be that you have a faulty database and a reseed is being triggered, or the index is bad. Identify the database in question and you can then move forward.

Sembee
  • 2,884
  • 1
  • 8
  • 11
  • I'll look in to this and update when I get more data the next time it happens. All I normally recall is seeing is that the queue from the /ecp page on the databases gets to a couple hundred/low thousand while transfers are happening. Thanks for the input, will update as soon as I can! – Abraxas Feb 14 '17 at 19:17