I think that the -- well, there is kind of a couple of things. One is that we're still seeing -- if we look at the 2 businesses, Andrew, I think it's worthwhile to look at records management versus data management, because data management, we still are seeing probably stronger headwinds in terms of getting to a, what I'll call, a stable floor than we are in records management. And the reason for that simply is that we're still finishing up the transition from tape being for backup and recovery to tape being an archival product. Right? So our storage volume, it continues to grow in the DM business, but how tape is being used is different. So as a result, we're doing this transition to a different service model, which happens to have a lower activity associated with it. And the thing I should highlight is, we are able to maintain our margins pretty closely during these drops in activity. So it's important to understand that it's an activity-driven. Whereas on the records management side, the data that we're looking at, we are further down that cycle, if I will, that transition to a more archival nature. So things like I was highlighting in healthcare, healthcare traditionally was a high-service business because it had a lot of, what we call, active file with medical records going backwards and forwards between the hospitals and our facilities. That obviously is gone away, or going away, fairly quickly in most locations. So I think in records management, that trend to archival nature of our services is further ahead, and we see that decline flattening out. So that's what we're really -- where we're looking at. At some point, it does get to the transition from active to archive you get through 80%, 90% of the transition, and I think that's where we're at with the records management. And we are -- I think we're well down that journey with data management, but we're not as far down the journey as we are with records management.
Andrew C. Steinerman - JP Morgan Chase & Co, Research Division: Great. And do you think we'll ever get to a point where the service revenue, activity-based service revenues, could be flat?