Performance and Reliability - IDEMA

Standardized Testing for SSD’s and HDDs
Performance and Reliability
IDEMA Technical Symposium
May 2008
Ed Dawson
Flexstar Technology
edawson@flexstar.com
www.flexstar.com

Agenda
Performance & Reliability Testing
• Why Test
• What is Standardized Testing
• Reliability & Performance Testing
• Device Specific Tests
• Storage Test Evolution

• Engineering Verification
– EVT DVT etc
– Development debug
• OEM Qualification
– Systems qualification
– Performance evaluation
– Four corner testing
– ORT
• Device Manufacturing
– Meet specifications
– Reliability / early life failures
– Fast failure analysis
• Field
– Failure analysis
– RMA confirmation
Why Test?

• Engineering Verification
– EVT DVT etc
– Development debug
• OEM Qualification
– Systems qualification
– Performance evaluation
– Four corner testing
– ORT
• Device Manufacturing
– Meet specifications
– Reliability / early life failures
– Fast failure analysis
• Field
– Failure analysis
– RMA confirmation
Why Standardized Tests?

1
Standardized Testing
Key Features of Standardized Testing
• Accuracy
• Repeatability
– Ability to reliably recreate same test
conditions globally
• Tight control over test conditions
– Environmental
– Test tool environment
• Scripting Based Tests
– Suits individual product requirements
– Easily sharable
• Ability to debug in place
– Interrogate / Trace failure conditions
– Remote testing/debugging
• Provides Data Collection & Correlation
– Consistency across the industry
– Understandable Reports
70,000.00
60,000.00
50,000.00
40,000.00
Rate (KB/s)
30,000.00
20,000.00
10,000.00
0.00
33
65
97
129
161
193
225
257
Serial Number VNRD2AC4CGVRHS
289
321
353
385
417
449
481
513
545
577
609
641
673
Test Points
705
737
769
801
833
865
897
929
961
993
Performance

Stop the questioning & finger pointing
by standardizing
The benefits of standardized, independent test systems
– Standard test hardware
– Unbiased and fair
– Repeatable test solutions
– Common test scripts
– Industry standard interfaces
– Correlation between
manufactures and EOM’s
– Speeds failure analysis
– Investment protection

Standards and Performance Testing
Storage Test Requirements
• Seek, transfer, & throughput performance
• Interface verification
• Specification compliance
• Head-disk interface integrity
• Media scans
• Voltage margins
• Long term reliability
• Environmental & Edge case

Performance Testing
General Performance Testing
– Seek & Transfer
• Sequential
• Random
• w/ & w/o compare
• etc, etc…
– Multiple streams
– Variable data stream width/ block
size, knee testing

Performance Testing
HDD Specific Performance Testing
– Radial Performance
– Seek mode performance
SSD SpecificPerformance
– Fragmented Wear Leveling

Reliability Testing
Design &
Manufacturing
Targeted test paths
Field &
Reliability

Reliability Testing
HDD design verification / qualification
( rotational media)
Root Cause
– Design/Components/DRAM
– Component Failures / i.e.
red phosphorous
– Head/Media instability
– Design margin
Test Process
⇒ Pattern reads/writes/verify
⇒ Power cycling
⇒ Extended test at temperature
⇒ Voltage margining
⇒ Temperature Cycling
⇒ Voltage margining & 4 corner

Reliability Testing
HDD variable field failure modes
(variable factors in rotational media)
Root Cause
– Contamination
– Handling, head laps
– Lube issues
– Motor bearings
– Growing defect lists
Test Process
⇒ Multiple full media R/W.
⇒ Head/Surface defects
⇒ W/R adjacent tracks
⇒ Dwell at elevated temperature

Reliability Testing
SSD design verification / qualification
– Endurance
Root Cause
– Performance-mfg. variability
– Bit failures / Data Retention
– Component Failures / i.e.
red phosphorous
– Write splice
– Metadata corruption
– Write performance
– Erase failures
– Design Margin
Test Process
⇒ Multiple writes
⇒ Performance verification
⇒ Disturb testing / pattern writes
⇒ Temperature Cycling
⇒ Voltage margining & 4 corner
⇒ Power cycling mid writes
⇒ Random I/O w/ power cycling
⇒ Cold write
⇒ Margined erase
⇒ Four Corners

Reliability Testing
SSD testing for variable failures
Root Cause
– Wear leveling performance
– Data Retention
– Reallocation errors
Test Process
⇒ Fragmentation tests
⇒ Temperature & power cycling
⇒ RW tests w/ power cycling
⇒ Write splice, cold writes

Variable Failure Modes
Post Manufacturing Failure Modes
HDD
• Contamination
• Handling
• Head slaps
• Lube Issues
• Motor Bearings
• Environmental
strains
ODD
• Media Failures
• Motor Issues
• Mechanical
• Optical pickup
SSD
• Endurance

AFR Predictions
Predicting a New ERA
• Historically, AFR cannot be correlated to manufacturing
or line integration failure rates.
• AFR will be more predictable in SSD, primarily due to the
eliminating of mechanical failures.
• Standardized front end testing for SSD's will lead to more
a more reliable predicted AFR benefiting storage
integrators.

SSD vs. HDD
What the Customer Expects
(in a world of tests developed by key OEMs)
• Complete functional equivalence and independence.
• Interface testing
• Reliability demonstration
• Performance demonstrations
• Endurance demonstration
• Power on / spin-up / sleep / resume from sleep
compliance.

Summary
What’s the role of organizations like
IDEMA?
At a point where test standards can be developed for
new solid state storage that are reflective of the
technology.

Thank You
Ed Dawson, Flexstar Technology