Would certainly consider a 'fewer bigger boxes, more hands-on' approach, BUT ONLY for clients that had good administration talent in-house and were already debu

Would certainly consider a 'fewer bigger boxes, more hands-on' approach, BUT ONLY for clients that had good administration talent in-house and were already debugging the managed services a lot anyway bc they're pushing the envelope. Managed services aren't perfect and I've seen them go bad, esp for things like failover. When they hiccup (either perf-wise or downtime) it's hard to debug effectively depending on what logs / stats the vendor makes available.