Most data centers have operating budgets for networking, storage, server, security and facilities departments.  Some extend this further with other departments like telco services, etc.  The budget silos that grow out of this model can wreak havoc on operating budgets.  The main problem here is communication.  Any new equipment or technology should be evaluated by all departments prior to coming into the space.  I have seen networking guys buy networking equipment that only later did the server team realize they would need new NIC cards to use.  I have seen equipment come in that there is no capacity to run only to create a race to find stranded power in the data center. 

The problem is relatively easy to solve if you think of all connections and the entire environment that a piece of equipment will need to operate.  For instance, take a server.  In general a server uses two network connections, two SAN connections, two power connections and will have a power /cooling requirement.  Figuring out the best place to implement the server for power and cooling should be the first priority.  Cabling can be used to reach switch ports if you aren’t using the short length limited DAC (Direct Attached Copper) for network.  There should be at least a spreadsheet to track power demands across the floor.  Floor loading is also important (although not often considered).   And of course there will also be a security requirement. 

Each server needs an address or two and a cables and cable management to keep from creating an air dam in the cabinet.  You can really think this way about any equipment that comes into the data center.  Now here is the trick- BEFORE ordering new stuff, talk to EVERY budget department in the data center and operations should be around too.  I can’t tell you how many data centers I go into where one department admits to knowing nothing about what another one is doing. 

There should be a control in place to force these discussions.  What if networking decides to put switches in the top of every cabinet even if about ¾ of the ports can’t be used.  Where is the CIO or company oversight here?  That certainly isn’t frugal, especially when there are better technologies out there that allow you to buy the number of switches you need based on the number of server ports not the number of cabinets.  And now all of these need power connections.  Who patrols the power whips and PDU’s?  It’s horrible to bring in new equipment only to find out there is no place to plug it in.

The best approach is the total involvement approach.  This will assure your ecosystem runs smoothly.  Each department should understand what it requires from the others and there should be a final fiscal signoff.