54
MANAGEMENTANDOPTIMIZATION
Lackofvisibility intoassets isa
recurring themewhendiscussing the
causesofoperational inefficiency. This
canbecorrectedbycreatingand
maintainingavisualmodelof thedata
centerenvironment that includes
equipment locationandspecifications.
Thismodeloften replacesmultiple
spreadsheets, consolidates information
fromvarioussources tostreamline
processesandenablesdatacenter
management toperformwhat-if
scenarios todetermine the impactof
changesbefore theyaremade.
Onceestablished, itbecomesa
foundationalmanagement tool that
drivesgreaterefficiency, agilityand
availability.
Equipmentmonitoringenablessome
manualprocesses tobeeliminatedand
isessential toachievinghigher
availability levels. Theability to receive
immediatenotificationofa failure—or
anevent thatcouldultimately lead toa
failure—throughacentralizedsystem
allows fora faster,moreeffective
response tosystemproblems.Equally
important, acentralizedalarm
management systemprovidesasingle
window intodatacenteroperations
thatprioritizesalarmsbycriticality,
ensuring themost serious incidents
receivepriorityattention.
Takenastep further, data from the
monitoringsystemcanbeused to
analyzeequipmentoperating trends
anddevelopmoreeffectivepreventive
maintenanceprograms.
Preventivemaintenance isanotherarea
wherepoorlydefinedorexecuted
processescanhavedisastrous
consequences.EmersonNetwork
Poweranalyzeddatacollectedby its
serviceorganization,whichmaintains
themostextensivedatabaseof service-
relatedevents for largeUPSsystems in
the industry, anddevelopeda
mathematicalmodel thatpredicts the
impactofpreventivemaintenanceon
UPS reliability. Thesecalculations
indicate that theUPSMeanTime
BetweenFailures (MTBF) forunits
that received twopreventiveservice
eventsayear is23 timeshigher than
amachinewithnopreventive
maintenanceserviceeventsper
year.
Monitoringandpreventive
maintenancecanhelpeliminate
someof the issues thatprevent
organizations fromaddressing the
causesof inefficiencies.Once these
preventivemeasuresare inplace,
themanagement teamcanbegin to
shift the focus fromshort-term issue
resolution to longer-termoptimization.
Forexample, thebestviewof ITpower
consumptioncomes from thepower
distributionunits inside racks. Theycan
provide receptacle-level visibility into
volts, kilowatts (kW), ampsandkWper
hour, providingadetailedviewofdata
centerenergyconsumption.When this
data isconsolidatedwithdata from
service-levelprocessorswithin the IT
equipment, itallowsunproductive
servers tobe identifiedand
decommissionedwhilesupporting
moredynamiccapacitymanagement.
Problems like resourceutilizationcan
begin tobeaddressed.
DCIMplatformssupport this typeof
activeoptimizationbydeliveringa
moreholisticviewofoperations. The
challenge is implementingasystem
thatcanhandle thehugeamountsof
datageneratedbydatacenter systems.
Thedatacenterappliancementioned
previously isdesignedspecifically to
collectdata fromavarietyofdevices
that share thesecritical characteristics:
•Real-time:
Historicaldatamaycome too late tobe
usefulormaynotaccurately reflect
currentconditions.AneffectiveDCIM
systemuses real-timeperformance
dataso thatmanagementcanknow
exactlywhat ishappeningatanygiven
timeandcanmodel the impactof
plannedchange tounderstand the
consequences immediately.
•Contextual:
It’snousecollectingdataacross
different typesofdevices if thatdata
can’tbe translatedandput intocontext
thatallowsperformanceacross
systems tobeunderstood.
•Prioritized:
Theproblemdatacentermanagers
face isn’t too little information. It is too
much.Datausedby themanagement
systemmustbeprioritized toensure
thatcriticaloperatingdata isnot
overwhelmedby less important
information.Bybringing together real-
time information fromacross thedata
center, putting it intocontextand
presenting inways that support
effectivedecisionmaking,DCIM frees
datacentermanagers frommanaging
blindandgives them theability tounify
ITand facilitiesdata toachieve:
- Higheravailabilityby recognizing
dependenciesandunderstand
performance in real time, predict the
impactofchangeandautomate
complexeventmanagementand
alarmnotifications.
- Betterefficiency througha
comprehensive inventoryofevery
asset’s flooror rackposition, role-
baseduser interfaces that simplify the
useofdetaileddataandaunified
viewof thedatacenter that facilitates
collaborativeplanning.
- Increasedutilization throughaccurate
insight intocurrentusage, real-time,
trendandhistorical change tracking
and theability topreview the impact
ofchangebefore it ismade.
TECHNOLOGY: INFRASTRUCTURE
THATCREATESTHEFOUNDATION
FORGROWTHANDAGILITY
Thedatacenter infrastructure—power
andcoolingsystems thatensuresafe
andcontinuousoperation—often
represents the leastagileandscalable
componentof thedatacenter. If
improperlydesignedandmaintained, it
canconstraingrowthandcontribute to
downtime.Conversely, the right
infrastructurecreatesa foundation for
continuousavailability, increased return
oncapital andcost-effectivegrowth.
STRATEGIESFORSCALABILITY
Heatdensityhasbeenoneof the top
threeconcerns identifiedby theDCUG
ineightof the lastnineyears.Yet,
according toDCUGsurveydata,
average rackpowerdensity in thedata
centerwasabout thesame in2013as in
2006,when thesurveywas initiated
(Figure4)
. This is likely the resultof the
7X24MAGAZINE SPRING2014
Figure4
