RAIDRAID

RAI

D

RAI

D

PLAYS ANIMPORTANTROLE IN SSDARRAYS

In data centers, RAID architecture has long played an essential role in protecting against hard disk drive (HDD) failures and enhancing performance.

With fast and reliable SSDs now becoming a key part of data center arrays, does RAID still play a role?

HDD ARRAYS REQUIRED HELPFROM RAID ARCHITECTURES FOR:

Improvedperformance

through greater parallelism

Enhanced reliability with error correction

and redundancy.

STATE-OF-THE-ART SSDs ADDRESS MANY OF THESE CHALLENGES

CRITICAL DATA STILL REQUIRES SYSTEM-LEVEL PROTECTION

EDUNDANT

RRAY OF

NDEPENDENT

RIVES

Writes distributed to alternating banks of �ash,reducing latency

LDPC manages �ash bit error rates and error

correctione�ciently

ENCODING

PCIe is faster and eliminates need for Host Bus Adapter

hardware

INTERFACING

For more insight, read the white paper:RAID Strategies in SSD Deployments

DOWNLOAD NOWSAMSUNG.COM/ENTERPRISESSD

Even though a single SSD is highly reliable, an array of

hundreds or thousands of SSDs in a data center encounters “hard” drive

failures eventually.

Redundancy – the ‘R’ in so�ware-based

RAID – protects againstpermanent system-wide

data loss when an entire drive fails.

STRIPINGSATA 2.0:

SATA 3.0:

SAS 3.0:

PCIe Gen 2:

PCIe Gen 3:

3.0 Gb/sec

6.0 Gb/sec

12.0 Gb/sec

x2: 8.0 Gb/sec

x4: 16.0 Gb/sec

x2: 15.75 Gb/sec

x4: 31.5 Gb/sec

PROS

Fastestrecovery time on rebuild

Leaste�cient

in capacity, may require 3-way

mirroring

RAID 10COMBINES

STRIPING WITHMIRRORING DATA

CONS

Best choice

where uptimeis most

important

More e�cientin capacity, lower cost

Longerrecovery

on rebuild, prone to

rebuild errors

RAID 5DISTRIBUTES A

PARITY VOLUME ACROSS ALL DRIVES

PROS CONS

Best mixof cost and

performance for most

SSD arrays

Protectsagainst 2simultaneousdrive failures

Morecomputationallyintense, longest

recovery

RAID 6ADDS A SECOND

DISTRIBUTED PARITYVOLUME FOR MORE

REDUNDANCY

PROS CONS

O�ersmore

protection for low

incremental cost