Preparing for a data center site survey
Posted: June 19, 2011 Filed under: Site Surveys Leave a commentBefore we ever start installing hardware or configuring software, we’re going to be conducting site surveys. Although preliminary surveys have already been done, we’ll be going a bit more in-depth to determine any last requirements for enterprise virtualization that are lacking. Initially, my thoughts were centered on the important things, like existing servers, storage, and networking assets. But a CISSP on our team created a document that goes much deeper and would catch more deficiencies that would impact a successful data center deployment. I’ll highlight the items of the likely less-thought-about issues that could hamper us if not properly accounted for:
Power
Distribution points, availability (how much downtime per day/week/month), how often main-line power is interrupted, are there brown-outs or line-spikes? Load factors for UPS units and current load (can the current UPS handle the hardware?), rack-mounted or facility UPS, UPS run-time under load, UPS battery life-cycle and maintenance, back-up generators, rack power distribution unit types and available receptacles with at least two per rack, each connected to different UPSs/distribution points/circuits
HVAC and Facility Air Handling
Capacity of chillers vs. heat output of equipment, is HVAC on UPS? Are there portable chillers in use? Rack flow capacity, rack row layout to include hot/cold rows, humidity
Fire Suppression
Type and location of suppressant, maintenance checklists, potential fire hazards
Physical Security
Types of access controls on entrances to data center (cipher, key, biometric, etc.), presence of camaras and capability (thermal, IR, pan, tilt, zoom), is building continuously occupied? Do people work in the data center? Distance of building location from hazards (near airport, near water but at lower level, prominent location, near high-traffic area), who has physical access? Verify accountability and access controls, emergency lighting during power-outage
Rack Configuration
Used/free RU space, power/signal grounding of racks, quality of existing wiring for power, networking, KVM, are racks lockable? Heat load/power draw for currently installed equipment, raised floor and capacity of floor, rack stability and anchoring
And I can’t help but include some of the storage, networking, and server notes, as well.
Storage
Makes/models in use, drive configuration and capacity, IP info, volume names, hostnames, domains, DR sites, IOps, quotas, interface names, 3rd party OS and apps, MAC/IP iSCSI addresses, protocols
Networking
Makes/models in use, used capacity, 10Gbps capable, IOS versions, management capability (WhatsUp Gold, SolarWinds, CNA), WAN connectivity between sites, topology and protocols, speed/latency, availability of fibre channel/fabric switching
Servers
Makes/models in use, hardware specs, 10Gbps capable, system performance baseline (CPU/memory/IO), OS versions, currently installed stand-alone ESX/ESXi, P2V viability, for virtual servers: ESX/ESXi version, location of .vmdk files, resource allocation/limits/reservations, ESX/ESXi performance baseline (CPU/memory/IO), attached storage details (LUNs, capacity, RAID levels, port assignments/WWNs), protocols supported, V2V viability