> Another problem with peering, zero redundancy. If you have half on > each server, one goes down, you lose half your boxes.
Not really. It just depends on how far you take the full-mesh model. If all of your data is clustered *and then* peered, no problem. 5-9's uptime SLAs all but require this setup (along with fully meshed switching and routing, etc.). Sandy ______________________________________________________________________ The HKSI-IMail Admin List is hosted by........ Humankind Systems, Inc. Questions, Comments or Complain like Hell.. mailto:[EMAIL PROTECTED] Message Archive... http://www.tallylist.com/archives/index.cfm/mlist.4 To Manage your Subscription......... http://humankindsystems.com/lists
