1. Synology RAID F1:
Rethinking RAID in all-Flash Environments
The standardizationof flashstorage hasbroughtwithithuge leapsin storage performance,
energyefficiency,anddevice size. Synologyremainsdedicatedtoprovidingadvancedsoftware RAID
solutionsthatmaximizereliabilitywhile adaptingtothe everchangingstorage landscape.
SolidState Drives and the Future of Storage
Cost pergigabyte has recently plummeted inflashmemory,makingpreviouslycostprohibitive
all-flashstorage arrays anewoption forsmall businesses.Ascompaniesbegintorelyonstorage pools
made entirelyof flashmemory,software solutionsrunningonappliancesthatsupportflash devices
mustalso adapt.Currently,most storage devicestreatSolidState Drives(SSDs) like HardDiskDrives
(HDDs),despite some crucial differencesinstructure andoperation. The importantdifferencehere is
that unlike HDDs,SSDshave a definedlimittothe numberof readand write operationstheycan
performintheirlifetime.
Hard Disk Failure:Redundancy is King
Anygroup of HDDs writtento withmirroredorsimilaramountsof data (seeninstorage devices
usingclassicRAIDconfigurations) will experience sporadicfailuresof memberdisks. Thiscanoccur from
environmental factors likevibration,moisture build-up,ordustcontamination.Because HDDsrelyon
movingpartsto read andwrite data, evenwhenenvironmentalfactorsare accountedfor,
manufacturingdefectsinthese movingpartscancause premature failure aswell. Thesefactors
combinedmake HDDfailuresmore random, and verydifficultto accuratelypredict.Since the lifespanof
an individualHDDina RAIDgroup cannotbe determined,itmakesthe mostsense totreatall member
disksas equally likelytofail.InclassicRAIDdeployments,dataiswritteninequal amountstoall member
disks.DiskredundancyinaRAID 1, 5, 6, etc.expectsa random,single (sometimes multiple) drivefailure,
and helpsprotectagainstdatalossinthose scenarios.
Currentsoftware RAIDsolutionsworkwell forHDDs,buttransitioningtoSSDsprovidesanew
challenge fordataprotectionsolutions.Withthe knowledge thatSSDshave a setnumberof read and
write operationsintheirlifespan,writingdataevenlyandconcurrentlytoall SSDsin an array could
resultinall of the drives “expiring”atthe same time!There isnoRAID technologyavailable thatcould
survive asimultaneousfailure of everymemberdisk,this scenariowouldbe adatalossdisaster. So,how
bestto approach thisnewissue facingall flasharrays?
SynologyRAID F1
SynologyRaidF1 isan algorithmdevelopedby Synologythathelpsmitigate this risk.RAIDF1isa
1-diskresiliencybasedonRAID5 concepts.RAID5 providesdataredundancyby“striping”dataand
paritybits (additional datausedforrecovery operations) acrossthree ormore memberdisks. Eachdrive
ina RAID5 groupstoressmall piecesof dataabout whatinformationisonthe otherdrivesinthatarray.
2. If a single drive fails,the data“about”the faileddrive storedonthe otherdisksisusedtorecreate the
informationfromthe faileddrive whenanotherisaddedtothe group.
RAID F1 differsfromRAID5 by selectingone SSDinthe array fordistributionof additional parity
bits,effectivelywritingmore datato one SSD than to othersinthe group. Writingmore data to a single
SSD ina groupof SSDsallowsfora single disktocomplete its estimatedlife cycle before anyothersin
the array. Thisprovidesasingle,predictable failure point,whichiswithinthe faulttolerance of the
array. The chosen SSD inthese casescould thenbe removedwhenitnearsthe endof its lifespan,and
replacedwithanewdrive.RAIDF1 will thenselectanew SSDon whichto write those extraparity
blocks,againchangingthe expected expiration rate of the newlyselected,andpreventingtotal array
failure due toconcurrentSSD expiration.
SynologyFS3017: Enterprise Hardware withoutthe Second Mortgage
RAID F1 isavailable onthe SynologyFS3017, an all flashstorage controllerwithdual Xeonsix
core processors,64GB of DDR4 ECC RAM(expandable to512GB), builtin10GbE (withsupportfor25GbE
and 40GbE), and comesequippedwiththe latestversionof DSM,Synology’sawardwinningNAS
operatingsystem.