User's Manual
Table Of Contents
- Understanding Silver Spring Networks Access Points
- Contents
- 1 About Access Points
- Overview
- Features
- AP Product Description
- Battery Backups
- Read Storage
- Power Requirements
- Rebooting or Recycling APs
- Standards Compliance
- Cellular Modem
- Addressing Schemes
- Specifications
- European Union Compliance
- Maintenance Procedures
- Surge Protection
- Redundant WAN Option
- Mounting Options
- WAN Options
- Throughput Performance
- Responding to a WAN Failure
- Multicast / Unicast Communications
- Networking
- Address Management
- Administration
- Configuration
- 2 FCC and Government Guidelines
- Index
Understanding Silver Spring Networks Access Points 1 About Access Points
Understanding Silver Spring Networks Access Points 4 March 2013 Silver Spring Networks 16
Responding to a WAN Failure
AnAPmightnotbereachablebecausethedevicecannotdialoutontheWAN.Thedevice
continuestotrydialingoutuntilitiseitherabletoconnect,orithasdialed308times.Ifitcan
connect,thedevicenotifiestheeventmanagementprogramthatithassuccessfully
recovered.
Ifitcannot,thenthemodemresetsitself.ThebehaviorofanAPrepeatedly
becomingunreachable,andthenrecoveringshortlyafter,iscalledflapping.
Asmallamountofflappingisnormal.Becauseadevicecanrecoveronitsown,waitforone
dayofdowntime,oronetotwodaysof
frequentflappingbeforeinvestigati ngfurther.
Filteringandreviewingtheeventmanagementprogrammailalertsregularlycanhelpto
identifydevicesthatneedattention.Thestandardoperatingprocedureregardingcollector
failure/managementisshowninFigure5onpage15.
Multicast / Unicast Communications
Commandssentfromhead‐endmonitoringapplications(NEM,GridScape,orHCM)
throughtheAccessPoint(AP)areunicast,asthereceiptofeachpacketmustbe
acknowledged.However,insimilarfashiontomulticast,commandscanbesenttoa
staticallyordynamicallydefinedgroupofRFdevicesforeaseofmanageability.
Inaddition,
theHESsends“Jobs”outinparalleltoAccessPoints,thattheninturn,sendcommandsout
toendpoints.Theresultingeffectisparallelprocessingthataccomplishesamulticast‐like
service.
Networking
EachNICinSilverSpringRFdevicesregisterswithtwoIPv6addresses–oneforeachoftwo
APs:aprimaryandsecondaryAP.Whenonedies,anotherisfound.Fromatopology
perspective,eachAPisitsownIPv6subnet.SoNICsareintwodifferentsubnets‐onefor
each
AP.
Theend‐to‐endprotocolwithintheNANisAES128/256bit,cryptoIPv6.AstheAccessPoint
understandsthefulltopologyoftheunderlyingmesh,theIPv6packetissourceroutedfrom
theAPtoeachenddevice–meaningthatthefullcommunicationpathfromtheAccessPoint
tothedeviceisspecified,hopbyhop,inthepacket.TheNANendpointtoAccessPoint
routingdecisionsaremadehop‐by‐hop(thatis,theNANendpointwillsendtheIPv6packet
toitsneighborthathastheleastcostroutetotheAccessPoint.Thenthisneighbor
will
forwardthepackettoitsneighborwiththebestrouteuntilthepacketreachestheAccess
Point.(TheAccessPointacquiresitsunderstandingofthefullmeshtopologybecauseit
sendsrouteadvertisementmessagesatrandomizedintervals.)
NANsystemonewaylatencyismeasuredat50msperhopbetweenany
pointsintheNAN.
TheSilverSpringnetworkdesignexpectsonaverage,nomorethan6hopstoanyendpoint
(suchasameterorDataLinkControl,DLC,device).Thiswouldtypicallyyield300msof
NANlatencytoanyRForDLCdevice.Inaddition,anAccessPointcan
processupto10
packetspersecond.EachDLCcommandconsistsofasinglepacket,withoneassociated
ACKpacket.WithnootherAMItraffictraversinganAccessPoint,anAccessPointcould
process100DLCcommandsandACKs,inapproximately30seconds.
Thecellularmodemhasstore‐and‐forwardGPSreportingcapabilitywithaccuracynoless
than10meters.