How do you balance the need for system upgrades with the necessity for minimal downtime at Harvard University?
