The paper makes it clear they're not counting servers waiting to be provisioned. But they're not so clear on how the tool makes the distinction, or how they kno

The paper makes it clear they're not counting servers waiting to be provisioned. But they're not so clear on how the tool makes the distinction, or how they know to exclude hot spares.