Two Extremes: Grid Computing and Globus vs. Massively parallel - - PowerPoint PPT Presentation

Two Extremes: Grid Computing and Globus vs. Massively parallel computing for big science

!"##"$%&'("##)&*+&,##-.-/& 0"12-3.1&.4&5621$7.89:&;$8$<21:&,=>$8-2=&?832<1$7.8&@1.A6& ,1<.882&B$7.8$#&C$D.1$3.1E:&C2$=219F"6&G.%6A78<&H$-"#"3E& & !"3F&9#"=2&$99"93$8-2&41.%I&J$K&L2M%A3FA&',BC):&;"1.8&C">8E'N!&O&;$="9.8):&J.D213& P$-.D&',BC):&Q.==&Q$8828D$A%&'N!R;$="9.8):&L$3F21"82&J"#2E',BC)&&

Administrivia, disclaimers, etc..

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

O! ?&$%&<."8<&3.&%287.8&>28=.19&$3&>$1".A9&6."839&"8&3F"9&6129283$7.8+&&QF$3&S"##&D2&4.1& .82&.4&3S.&12$9.89I&

Y! ?3&"9&$&>28=.1&S2&A92&$8=&?&$%&=29-1"D"8<&.A1&"841$931A-3A12& Y! ?3&"9&$&>28=.1&?&$%&$S$12&.4&"8&$&96$-2&3F$3&922%2=&3.&D2&.4&"8321293&3.&E.A+&

O! QF212&$12&3S.&>21E&<..=&12$9.89&3F$3&3F292&$12&8.3&28=.192%2839&

Y! ,1<.882&=.29&8.3&621%"3&"3&$8=&?&#"/2&%E&K.D&"& Y! !F$3&S.1/9&.1&"9&<..=&"9&-.%6#232#E&9"3A$7.8&=2628=283&$8=&"3&S.A#=&D2&4..#"9F&.4&%2&3.&31E& $8=&28=.192&.82&.4&3F292&61.=A-39&8.3&/8.S"8<&$8E3F"8<&$D.A3&E.A1&9"3A$7.8+&

GF"-$<.&N8"4.1%&;2278<:&Z[&H2D:&Z\]Z& Z&

Argonne National Laboratory (ANL)

!! ,1<.882^9&%"99".8&"9&3.&$66#E&$&A8"TA2&%"_&.4&S.1#=R-#$99&9-"28-2:&28<"8221"8<&$8=&A921& 4$-"#"729&3.&=2#">21&"88.>$7>2&1292$1-F&$8=&32-F8.#.<"29+&!2&-12$32&82S&/8.S#2=<2&3F$3& $==129929&3F2&%.93&"%6.13$83&9-"287`-&$8=&9.-"23$#&822=9&.4&.A1&8$7.8+& !! ,1<.882^9&93128<3F9&"8-#A=2&

O! *821<E&V3.1$<2&'GF2>E&a.#3:&C"&?.8&32-F8.#.<E)& O! ,#3218$7>2&*821<E&'V.#$1:&!"8=:&D".4A2#9:&4A2#&$8=&2821<E&2b-"28-E)& O! BA-#2$1&*821<E&'V$42&12$-3.1&32-F8.#.<E:&4A2#&-E-#2&%$8$<2%283)& O! (".#.<"-$#&$8=&*8>"1.8%283$#&9E932%9&'<28.%"-9:&-#"%$32)& O! B$7.8$#&V2-A1"3E&'4.1289"-9:&=2-"9".8&9-"28-29:&9289.19:&%$321"$#9:&-ED21&92-A1"3E)& O! V-"287`-&N921&H$-"#"729&

Y! ,=>$8-2=&XF.3.8&V.A1-2&',XV)& Y! G28321&4.1&B$8.9-$#2&;$321"$#9&'GB;)& Y! ,1<.882&Q$8=2%&C"8$-&,--2#21$3.1&VE932%&',QC,V)& Y! *#2-31.8&;"-1.9-.6E&G28321&'*;G)& Y! Q1$896.13$7.8&J292$1-F&$8=&,8$#E9"9&G.%6A78<&G28321&'QJ,GG)& Y! ,3%.96F21"-&J$="$7.8&;2$9A12%283&G#"%$32&J292$1-F&H$-"#"3E&',J;)& Y! C2$=219F"6&G.%6A78<&H$-"#"3E&'CGH&R&3F$3^9&%E&=">"9".8)&

!! FU6Icc28+S"/"62="$+.1<cS"/"cN8"32=dV3$329d026$13%283d.4d*821<EdB$7.8$#dC$D.1$3.1"29&

GF"-$<.&N8"4.1%&;2278<:&Z[&H2D:&Z\]Z& e&

Leadership Computing Facility

!! QF2&,CGH^9&%"99".8&"9&3.&$--2#21$32&%$K.1&9-"287`-&="9-.>21"29&$8=&28<"8221"8<& D12$/3F1.A<F9&4.1&FA%$8"3E&DE&=29"<8"8<&$8=&61.>"="8<&S.1#=R#2$="8<&

• .%6A78<&4$-"#"729&"8&6$138219F"6&S"3F&3F2&-.%6A3$7.8$#&9-"28-2&-.%%A8"3E+&

& !2&931">2&3.&D2&3F2&4.1241.83&-.%6A3$7.8$#&-28321&4.1&2_328="8<&3F2&41.87219&.4& 9-"28-2&DE&9.#>"8<&/2E&61.D#2%9&4.1&3F2&8$7.8&3F$3&12TA"12&"88.>$7>2&$661.$-F29& $8=&3F2&#$1<293R9-$#2&9E932%9+& !! GA11283#E&

O! ?83126"=I&ffg&QHC5X&(#A2&@282cX:&h\L&8.=29:&]i\L&-.129:&[\&Q(&J,;:&[X(&.4&="9/:&']\)& f]Z&6.13&]\@&9S"3-F29:&$661._"%$32#E&e\\&.3F21&921>219+&

!! 026#.E"8<&3F"9&E2$1&

O! ;"1$I&]\XHC5X&(#A2&@282cj:&h[L&8.=29:&gi[L&-.129:&gi[&Q(&J,;:&Z[X(&="9/:&'h)&ih[& 6.13&h\@&?8`8"D$8=&9S"3-F29:&]Z\k&.3F21&921>219&'`#2&921>219&$12&8.S&2%D2==2=&"8&3F2& 93.1$<2)+&

!! 5A1&4$-"#"3E&"9&.628&3.&KA93&$D.A3&$8E.82&'2_-263&3F2&lV3$32&V6.89.19&.4& Q211.1"9%m)+&&QF"9&"8-#A=29&-.%%21-"$#&-.%6$8"29&'X1$U&$8=&!F"382E:&X1.-3.1&$8=& @$%D#2:&@*&F$>2&$##&1A8&.8&.A1&%$-F"82)+&&QF212&"9&8.&-F$1<2+&&?3&"9&<">28&$9&$&<1$83& 3.&3F2&D293&9-"28-2&S"3F&3F2&<12$3293&"%6$-3+&&?8=A931E&.A312$-F&"9&$&4.-A9&1"<F3& 8.S+&&Q$#/&3.&%2&.n"82&"4&E.A&$12&"83212932=+&

GF"-$<.&N8"4.1%&;2278<:&Z[&H2D:&Z\]Z& h&

Two Extremes… Both are big, but moving them would require very different solutions…

!! ;$8E&9%$##&3F"8<9& !! QF2&V$F$1$&O&[_]\Zg&<1$"89&.4&9$8=& !! @..<#2&O&3289&.4&%"##".89&.4&TA21"29&$& =$E& !! QF292&61.D#2%9&12TA"12&#"U#2&.1&8.&

• ..1="8$7.8&'-.%%A8"-$7.8)+&

!! QF"9&"9&@1"=&c&o"<F&QF1.A<F6A3& G.%6A78<&'oQG)& !! 582&FA<2&3F"8<& !! ,E219&J.-/&O&]&1.-/&S2"<F"8<&h;&3.89& !! jA$83A%&GF1.%.=E8$%"-9&'jG0)&O& ]\\9&.4&%"##".89&.4&-.12&F.A19&.8&.82& 61.D#2%+& !! QF292&61.D#2%9&$12&D.A8=2=&DE&

• ..1="8$7.8&'-.%%A8"-$7.8)+&

!! QF"9&"9&%$99">2#E&6$1$##2#&-.%6A78<&

GF"-$<.&N8"4.1%&;2278<:&Z[&H2D:&Z\]Z& f&

Grid Computing and Globus

So what is Grid Computing?

!! QF2&612-"92&=2`8"7.8&S$9&#$"=&.A3&"8&$&6$621&-$##2=&lQF2&,8$3.%E&.4&3F2&@1"=m&DE& ?$8&H.9321&',BC):&G$1#&L2992#%$8&'NVGc?V?):&V32>2&QA2-/2&',BC)+&

O! p2_"D#2:&92-A12:&-..1="8$32=&129.A1-2&9F$1"8<&$%.8<&=E8$%"-&-.##2-7.89&.4&"8=">"=A$#9:& "8973A7.89:&$8=&129.A1-29qSF$3&S2&12421&3.&$9&>"13A$#&.1<$8"r$7.89+&

!! !F$3&"893$87$7.8&S$9&-F.928s&

O! @1"=&G.%6A78<&"9&$&61.3.-.#&c&921>"-2&-2831"-&>"2S&.4&3F2&S.1#=+&&QF2&"=2$&S$9&3.&-12$32& $&93$8=$1=&923&.4&61.3.-.#9&$8=&921>"-29&$8=&92-.8=$1"#E:&"83214$-29&3.&3F2%:&9.&3F$3& 3F212&S$9&$&A8"4.1%&S$E&3.&$--299&$8E&129.A1-2&3F$3&9A66.132=&3F2%+& O! QF"9&"9&>21E&9"%"#$1&3.&3F2&?8321823&%.=2#&'?X:&QGX:&oQQX)&$8=&"8&-.831$93&3.&3F"8<9&9A-F&$9& G#.A=&G.%6A78<&'?841$931A-3A12&$9&$&V21>"-2:&?$$V):&V-"28-2&X.13$#9&'V.tS$12&$9&$& V21>"-2:&V$$V):&23-++&&o.S2>21:&3F2E&$12&8.3&%A3A$##E&2_-#A9">2+&&u.A&-.A#=&1A8&E.A1&@1"=& 921>"-29&.8&"841$931A-3A12&96A8&A6&"8&$&-#.A=&'3F212&$12&-.%6$8"29&3F$3&=.&3F"9)&

!! !F212&$12&S2&3.=$Es&

O! ?4&E.A&12$=&3F2&6$621:&%.93&.4&SF$3&"9&1242112=&3.&$9&3F2&J29.A1-2&C$E21&S$9&-.%6#232=& $8=&"9&SF$3&"9&"8&A92&3.=$E&'?&S"##&<">2&%.12&=23$"#9)+&&QF2&82_3&#2>2#&A6:&1242112=&3.&$9&3F2&

• .##2-7>2&#$E21:&F$=&%$8E&$U2%639:&`39:&$8=&93$139:&DA3&.8#E&$&-.A6#2&.4&3F"8<9&2>21&

12$##E&<$"82=&31$-7.8+&

!! ,8E.82&-.89"=21"8<&A9"8<&@1"=&G.%6A78<&9F.A#=&=2`8"32#E&12$=&3F2&,8$3.%E&.4& 3F2&@1"=+&&QF2&=23$"#9&.4&@#.DA9&F$>2&-F$8<2=:&DA3&%A-F&.4&"3&97##&$66#"29+&

O! FU6IccSSS+<#.DA9+.1<c$##"$8-2c6AD#"-$7.89c6$6219c$8$3.%E+6=4&

GF"-$<.&N8"4.1%&;2278<:&Z[&H2D:&Z\]Z& g&

Batch Computing

!! ?4&2>21E.82&"8&3F2&1..%&"9&4$%"#"$1&S"3F&3F"9:&#23&%2&/8.S&$8=&?&S"##&9/"6&3F"9&9#"=2:& DA3&@1"=&c&V-"287`-&c&o"<F&X214.1%$8-2&G.%6A78<&"8&<2821$#&"9&D$92=&.8&D$3-F&

• .%6A78<&$8=&%A-F&.4&SF$3&?&9$E&S.8v3&%$/2&92892&.3F21S"92+&

!! ($3-F&>9+&?8321$-7>2&

O! D$3-F&]&&'D-F)!"#

Y! !"#$%&'()*+#,-.*!-*&,&923&.4&=$3$&.1&K.D9&3.&D2&61.-2992=&"8&$&9"8<#2&61.<1$%&1A8+& Y! )+"/"&$%&'()*:&$%&'(+,-.:&$%&'(+)/&Q.&$992%D#2&.1&61.-299&$9&$&D$3-F+&

O! "8w321w$-w7>2&&'831R/3>)012"#

Y! 0"#,-78<&.1&-$6$D#2&.4&$-78<&.8&2$-F&.3F21+& Y! 1"#$%&'()*+#,-.*!-*&54&.1&12#$78<&3.&$&61.<1$%&3F$3&1296.8=9&3.&A921&$-7>"3E+&

!! ,9&$&%23$6F.1&3F"9&"9&$&9"3R=.S8&1293$A1$83&>9+&$&-$42321"$&

O! ?8&3F2&9"3R=.S8&1293$A1$83&E.A&6#$-2&E.A1&.1=21:&S$"3&$&SF"#2:&$8=&"3&"9&D1.A<F3&3.&E.A+& O! ?8&3F2&-$42321"$:&"3&"9&.8&=2%$8=+&&B.&S$"78<+&

!! QF2&"=2$&"9&3F$3&E.A&F$>2&$&>$#A$D#2&129.A1-2&3F$3&"9&A8=21R61.>"9".82=&.1&"8&

• .83287.8&'-.%6A32&-#A9321)+&&&

O! Q.&<23&$--299&3.&3F"9&3.&129.A1-2:&E.A&2_2-A32&$&-.%%$8=&3F$3&32##9&3F2&9-F2=A#21&F.S& %$8E&8.=29&E.A&S$83:&SF$3&2_2-A3$D#2&3.&1A8:&4.1&F.S&#.8<:&23-&'E.A&6#$-2&E.A1&.1=21)+&

Y! T9AD&R3&Z\&R8&]\Zh&cD"8c9#226&i\\&

O! QF2&9-F2=A#21&3F28&6#$E9&$&D"<&<$%2&.4&Q231"9&$8=&1A89&3F2&K.D9&$9&.67%$##E&$9&6.99"D#2& $--.1="8<&3.&#.-$#&6.#"-E&'E.A1&.1=21&"9&6126$12=&$8=&D1.A<F3&3.&E.A)&

GF"-$<.&N8"4.1%&;2278<:&Z[&H2D:&Z\]Z& [&

So, what is Globus?

!! ?3&"9&$&61.K2-3&$8=&$9&S"3F&%.93&9A--2994A#&61.K2-39:&"3&93$132=&.>21&$&D221&"&

O! ?$8&H.9321&$8=&G$1#&L2992#%$8&S212&"8&V$8&H1$8-"9-.&$8=&="9-A992=&F.S&3F2E&-.A#=& %21<2&3F2"1&-.%6#"%283$1E&1292$1-F&61.K2-39+&&&

!! C"/2&$8E&#$1<2:&#.8<&#">2=&61.K2-3&'$D.A3&]f&E2$19&8.S)&"3&F$9&-F$8<2=&.>21&7%2+& !! ?3&"9&$#9.&$&!""#$%!&4.1&28$D#"8<&@1"=&G.%6A78<& !! QF2&3F"8<9&3F$3&S.A#=&#"/2#E&%$U21&3.&E.A&3.=$E&$12I&

O! V2-A1"3E&

Y! @1"=&V2-A1"3E&?841$931A-3A12&'@V?)&O&W+f\x&-217`-$329&S"3F&=2#2<$7.8&

O! P.D&VAD%"99".8&

Y! @1"=&J29.A1-2&,##.-$7.8&$8=&;$8$<2%283&'@J,;)&R&02>2#.6&$&93$8=$1=&K.D&9AD%"99".8& "83214$-2&$8=&3F28&S1"32&l9F"%9m&3.&31$89#$32&3.&3F2&>$1".A9&9-F2=A#219&.A3&3F212&'X(V:&CVH:& C.$=C2>2#21:&;$A":&G.D$#3:&23-)&

O! 0$3$&;$8$<2%283&

Y! @1"=HQX&R&92-A12:&F"<F&9622=:&$8=&12#"$D#2&%.>2%283&.4&#$1<2&=$3$9239& Y! @#.DA9&58#"82&'@5)&O&V$9V&.y21"8<&.4&$&F"<F#E&12#"$D#2&=$3$&31$89421&921>"-2& O! @1"=HQX&S$9&12#"$D#2&A8#299&3F2&-#"283&4$"#2=&"8&SF"-F&-$92&$##&93$32&S$9&#.93+&&@5& $==129929&3F$3+&

O! QF2&*-.9E932%&3F$3&F$9&=2>2#.62=&$1.A8=&"3&

Y! ;E61._E&'BGV,)&O&8.S&-.%29&S"3F&3F2&3..#/"3z&&$##.S9&%.12&-2831$#"r2=&-.831.#&.4&W+f\x&

• 217`-$329+&

Y! !.1/p.S&%$8$<2%283&O&VS"t&'NG):&X2<$9A9&'NVGc?V?):&$8=&6$17-A#$1#E&G.8=.1&'N!R;$="9.8)&

GF"-$<.&N8"4.1%&;2278<:&Z[&H2D:&Z\]Z& x&

Open Science Grid (OSG) HTC at the National Level

GF"-$<.&N8"4.1%&;2278<:&Z[&H2D:&Z\]Z& ]\&

Large Hadron Collider

Chicago Uniform Meeting, 28 Feb, 2012

]]&

The LIGO Project

!! C$921&?8321421.%2321&@1$>"3$7.8$#&!$>2&5D921>$3.1E& !! C?@5&"8931A%2839&"8&!$9F"8<3.8&V3$32&$8=&C.A"9"$8$& !! 0A1"8<&9-"28-2&1A89:&61.=A-2&A6&3.&Z&321$DE329&621&=$E& !! XAD#"9F2=&$#.8<&S"3F&%23$=$3$&$3&G$#32-F&'$1-F">$#&9"32)& !! J26#"-$32=&$3&A6&3.&]\&.3F21&C?@5&9"329&

O! C?@5&9-"287939&3E6"-$##E&%.>2&=$3$&9239&82$1&3.&-.%6A3$7.8$#&-#A93219&$3& 3F2"1&9729&

!! QF2&C?@5&0$3$&@1"=&

GF"-$<.&N8"4.1%&;2278<:&Z[&H2D:&Z\]Z& ]Z&

Earth System Grid Federation (ESGF)

234%'&# 5$6)'78)/##

!! ;223&3F2&962-"`-&822=9&.4&8$7.8$#&$8=&"83218$7.8$#&-#"%$32&projects for distributed databases, data access, and data movement. !! Provide a wide range of climate data-analysis tools and diagnostic methods to international climate centers and U.S. government agencies. !! 2-&).9%&)#,-#%#':;;%$:9%78)#49:$;)3</:;8,-.#)-8,9:-3)-&#(,.(;=# 4>$;,',?)*#';,3%&)#*%&%#/)&/#>/,-.#*,/&9,$>&)*#/&:9%.)#3%-%.)3)-&@# 9)3:&)#(,.(<4)9A:93%-')#>-,&/@#(,.(<$%-*B,*&(#B,*)<%9)%#-)&B:9C/@# %-*#>/)9#*)/C&:4#4;%D:93/"#

!! Massive data archives (PB moving to XB) !! Multiple data centers worldwide

• !

Existing IT infrastructure and separate security domains !! Heterogeneous data sources (models, observations, reanalysis) !! Multiple physical realms (atmosphere, ocean, land, sea ice) !! Multiple data, metadata formats and conventions !! Multiple scales (global, regional and local) !! Cyber security !! Multiple audiences (scientists, policy makers, students, educators)

The ESGF consortium’s mission is to provide climate researchers worldwide with a system of federated science gateways to access data, information, models, analysis tools, and computational capabilities required to evaluate ultrascale data sets. Its goals are to make data more useful to climate researchers by developing collaborative technology that enhances data usability, and provide a universal and secure Web-based data access portal for broad-based multi-model data collections.

E)4;:=3)-&#:A#F;,3%&)#E%&%#G%-%.)3)-&#%-*#H'')// #

Contact Dean N. Williams (williams13@llnl.gov) for more information

GF"-$<.&N8"4.1%&;2278<:&Z[&H2D:&Z\]Z& ]e&

Earth System Grid

GF"-$<.&N8"4.1%&;2278<:&Z[&H2D:&Z\]Z& ]h&

Main ESG Portal IPCC AR4 ESG Portal 146 TB of data at four locations

!! 1,059 datasets !! 958,072 files !! Includes the past 6 years of joint DOE/NSF

climate modeling experiments

35 TB of data at one location

!! 77,400 files !! Generated by a modeling campaign coordinated by the

Intergovernmental Panel on Climate Change

!! Model data from 13 countries

4,910 registered users 1,245 registered analysis projects Downloads to date

!! 30 TB !! 106,572 files

Downloads to date

!! 245 TB !! 914,400 files !! 500 GB/day

(average)

> 300 scientific papers published to date based on analysis of IPCC AR4 data

ESG facts and figures

Worldwide ESG user base

IPCC Daily Downloads (through 7/2/07)

V#"=2&G.A1329E&.4&0$>2&(218F.#=3:&5JBC&

GF"-$<.&N8"4.1%&;2278<:&Z[&H2D:&Z\]Z& ]f&

Advanced Photon Source

Chicago Uniform Meeting, 28 Feb, 2012

APS Beam line APS HPC APS DMZ Public Server

Beam line Controls Data Acquisition

Beam line Storage GridFTP Server Infiniband

Lustre Parallel FS

GridFTP Server APS Tier2 Firewall ANL Tier1 Firewall Public Network GridFTP Server Myproxy Server

Beam line controls pushes data to HPC

User gets credentials from Myproxy server User pulls data from public GridFTP server

]i&

Dark Energy Survey

*_312%2#E&12=R9289"7>2& f\\&%2<$6"_2#&-$%21$& 0$3$&$-TA"9"7.8&9E932%&4$93&28.A<F& 3.&3$/2&"%$<29&"8&]g9& Q2#29-.62&

GF"-$<.&N8"4.1%&;2278<:&Z[&H2D:&Z\]Z& ]g&

Dark Energy Survey

!! QF2&0$1/&*821<E&VA1>2E&'0*V)&S"##&F$>2&1.A<F#E&fZf& .D921>"8<&8"<F39&.>21&$&`>2RE2$1&9A1>2E&621".=& 'Z\]]RZ\]i)& !! *$-F&.4&3F292&8"<F39&S"##&61.=A-2&$661._"%$32#E&e\\& @(&.4&1$S&=$3$&$8=&Z+f&Q(&3.3$#&=$3$&61.=A-39&$t21& 61.-299"8<&& !! !"3F"8&2$-F&.4&"39&12-283&0$3$&GF$##28<29:&3F2&0$3$& *821<E&VA1>2E&0$3$&;$8$<2%283&'0*V0;)&32$%&F$9& 61.-2992=&]\&9A-F&8"<F39&.4&9"%A#$32=&0*V&=$3$:& 61.=A-"8<&ZfRe\&Q(9&=$3$&

GF"-$<.&N8"4.1%&;2278<:&Z[&H2D:&Z\]Z& ][&

Let’s get down to brass tacks

!! H1.%&$&DA9"8299&621962-7>2:&SF$3&S.A#=&?&9A<<293&E.A&3$/2&$&F$1=&#../&$3I& !! ?4&#$1<2&=$3$&%.>2%283&"9&$&61.D#2%&4.1&E.A:&3F28&#../&$3&@1"=HQX&$8=&@#.DA9& 58#"82&

O! @1"=HQX&"9&%$3A12z&&?3&F$9&D228&$1.A8=&4.1&]\&E2$19&$8=&3F2&9"329&3F$3&126.13&3.&A9& $>21$<2&$D.A3&]f;&.621$7.89&621&=$E&S.1#=&S"=2+&&?3&"9&$>$"#$D#2&"8&92>21$#&.4&3F2&C"8A_& ="931"DA7.89+&

Y! V2-A1"3E&"9&2"3F21&>"$&@V?&'W+f\x&61._"29)&.1&99F&

O! @#.DA9&58#"82z&&?3&"9&$&12#$7>2&82S&-.%21:&DA3&"3&9.#>29&$&12$#&61.D#2%:&61.>"=29&$& 92-A1"3E&9.#A7.8&3F$3&%.93&DA9"829929&-$8&S.1/&S"3F:&"3&1A89&.8&,%$r.8&9.&3F2& "841$931A-3A12&"9&612UE&12#"$D#2&$8=&9.&4$1:&3F2&9.tS$12&F$9&D228&612UE&9.#"=+&

Y! V7##&A929&W+f\x:&DA3&$A3F287-$7.8&3.&3F2&#.-$#&%E61._E&921>21&3.&<23&3F2&61._E&"9&>"$&SF$32>21& #.-$#&92-A1"3E&E.A&F$>2&"8&6#$-2&'X,;)+&

!! ?4&o"<F&QF1.A<F6A3&G.%6A78<&c&GE-#2&9-$>28<"8<&"9&.4&"8321293&3F28&#../&$3&G.8=.1&

O! B.3&6$13&.4&@#.DA9:&DA3&$&6$13821&.4&@#.DA9+&&G$8&"83214$-2&3.&9"329&S"3F&@J,;&1A88"8<:& DA3&@J,;&"9&8.3&12TA"12=& O! a21E&%$3A12:&$>$"#$D#2&"8&$3&#2$93&J2=o$3:&F$9&9.%2&9A--2994A#&-.%%21-"$#&A9219+&

GF"-$<.&N8"4.1%&;2278<:&Z[&H2D:&Z\]Z& ]x&

!! J2#"$D#2&`#2&31$89421+&

O! *$9E&l`12R$8=R4.1<23m&31$894219& O! ,A3.%$7-&4$A#3&12-.>21E& O! o"<F&6214.1%$8-2& O! ,-1.99&%A#76#2&92-A1"3E&=.%$"89&

!! B.&?Q&12TA"12=+&

O! V.tS$12&$9&$&V21>"-2&'V$$V)&

Y! B.&-#"283&9.tS$12&"893$##$7.8& Y! B2S&42$3A129&$A3.%$7-$##E&$>$"#$D#2&

O! G.89.#"=$32=&9A66.13&{&31.AD#29F..78<& O! !.1/9&S"3F&2_"978<&@1"=HQX&921>219& O! @#.DA9&G.882-3&9.#>29&l#$93&%"#2&61.D#2%m&

!! ,&12-.%%28=2=&921>"-2&.4&WV*0*:&B*JVG:&*V823:&NL& B@V:&NGC,:&N+&;"-F"<$8:&|& What is Globus Online?

GF"-$<.&N8"4.1%&;2278<:&Z[&H2D:&Z\]Z& Z\&

… with MyProxy OAuth (new feature)

& @#.DA9&58#"82& & ;EX1._E& 58#"82&G,&

Username password

X,;& C.-$#&,A3F287-$7.8&VE932%& 'C0,X:&J,0?NV:&B?V:&L21D21.9&23-)& @1"=HQX& V21>21&

certifficat e Q1$89421&& 12TA293& certificate

I&)4#1# V326&[& V326&x&

C.-$#& V3.1$<2& @1"=HQX& V21>21&

certificate Access files

V326&]\& V326&]]&

,A3F287-$7.8&&& {&0$3$&Q1$89421&& ,A3F.1"r$7.8&

5$A3F& V21>21&

Username password Username password certificate certificate Redirect

I&)4#J# V326&h& V326&f& V326&i& V326&g& V326&]&

,--299&*8=6."83&

GF"-$<.&N8"4.1%&;2278<:&Z[&H2D:&Z\]Z& Z]&

Globus Online in Action

28.6 Terabytes 31,000 files 56h 44m No human involvement Astrophysics simulation data generated in Tennessee, moved to Illinois for visualization (Enzo, UCSD; Futures Lab, Argonne)

GF"-$<.&N8"4.1%&;2278<:&Z[&H2D:&Z\]Z& ZZ&

@G;N&

& @#.DA9&58#"82& & ;EX1._E& 58#"82&G,&

Username password

X,;& C.-$#&,A3F287-$7.8&VE932%& 'C0,X:&J,0?NV:&B?V:&L21D21.9&23-)&

Username password certificate QCV&F$8=9F$/2& certifficate Q1$89421&& 12TA293& certificate

V326&e&

V326&h&

V326&f& V326&i&

V326&g&

C.-$#& V3.1$<2& @1"=HQX& V21>21&

certificate Access files V326&[&

V326&x&

,A3F287-$7.8&&& {&0$3$&Q1$89421&& ,A3F.1"r$7.8&

V326&]&

,--299&*8=6."83& Username password

V326&Z&

@1"=HQX& V21>21&

Globus Connect Multi User

GF"-$<.&N8"4.1%&;2278<:&Z[&H2D:&Z\]Z& Ze&

High Throughput Computing

!! (5?BG&'FU6IccD."8-+D21/2#2E+2=Ac&)&

O! @12S&.A3&.4&3F2&V*Q?&$3&o.%2&61.K2-3& O! ?841$931A-3A12&3.&923&A6&$8E&l$3&F.%2m&61.K2-3& O!

• $9&32$%9&$8=&-.%6277.89&SF"-F&28-.A1$<29&6$17-"6$7.8&

O! B22=&3.&1A8&2$-F&K.D&%A#76#2&7%29&3.&F$>2&-.8`=28-2&"8&3F2&129A#39&

!! ?832#&'?83218$#&02>2#.6%283&X1.K2-3)&

O! *>21E&8.8R%.D"#2&%$-F"82&"8&3F2&-.%6$8E&S$9&$>$"#$D#2& O! ?&F$>2&F2$1=&-#$"%9&.4&F"<F&[\v9&4.1&A7#"r$7.8& O! H.##.S2=&3F2&="A18$#&-E-#2&

Y! V"8<$6.12&1$8&-F"6&=29"<8&>21"`-$7.8&-.=29&.8&92-123$1"29&%$-F"82&"8&G$#"4.18"$&$3&8"<F3+&

!! G.8=.1&'FU6Icc1292$1-F+-9+S"9-+2=Ac-.8=.1c&)&

O! 029"<82=&4.1&#$1<2&']\}f)&3$9/9&$8=&#$1<2&']\}h)&129.A1-29& O!

• "<F#E&129"#"283&

O! ;$3-F%$/"8<&c&G#$99&,=>21792%2839&

Y! J29.A1-29&$=>21792&3F2"1&-$6$D"#"729:&K.D9&$=>21792&3F2"1&822=9:&-.8=.1&$11$8<29&$&%$11"$<2&%$=2&"8&9"#"-.8&"& Y! B.&.82&2#92&3F$3&?&$%&$S$12&.4&F$9&$&<..=&9E932%&4.1&3F"9+&

O! @#"=2R"8v9&&

Y! QF2&K.D&E.A&93$13&.8&3F2&-#A9321&"9&G.8=.1&93$13=:&SF"-F&12<"93219&"392#4&$8=&3F28&D2-.%29&$&129.A1-2+& Y! FU6IccSSS+A9-%9+.1<cV.tS$12G.%6A78<c@1"=c!;Vc<#"=2"8!;Vc=.-+61=c"8=2_+F3%#&

GF"-$<.&N8"4.1%&;2278<:&Z[&H2D:&Z\]Z& Zh&

Zf&

Condor Pool Layout: running a job

= Communication Pathway = Process Spawned

Submit Host Master schedd

shadow shadow

Execute Host Master startd

starter starter Job Job

Execute Host Master startd

starter Job shadow

GF"-$<.&N8"4.1%&;2278<:&Z[&H2D:&Z\]Z&

Dynamically deployed on grid resources

GF"-$<.&N8"4.1%&;2278<:&Z[&H2D:&Z\]Z& Zi&

A Cool Condor Example

!! ,&6F$1%$-2A7-$#9&-.%6$8E&822=2=&3.&1A8&]\:\\\&K.D9+& O! ?3&S$9&297%$32=&3.&F$>2&3$/28&S22/9:&.1&6.99"D#E&2>28&%.83F9&.8&3F2"1& 2_"978<&"841$931A-3A12+& O! oQG&c&@1"=&"9&.t28&$&<..=&%$3-F&4.1&("<&XF$1%$&=A2&3.&3F2"1&822=&4.1& 6$1$%2321&9S2269+& !! H.1&$&61..4&.4&-.8-263:&3F2E&32$%2=&S"3F& O! GE-#2&G.%6A78<&'31"29&3.&F"=2&3F2&-.%6#2_"3E)& O! G.8=.1&'3.&F$8=#2&9-F2=A#"8<:&129"#"28-E)& O! ,%$r.8&'3.&<23&3F2&$-3A$#&"841$931A-3A12)& !! ]ZIef&O&]\:\\\&P.D9&9AD%"U2=&$8=&12TA2939&4.1&D$3-F29&.4&-.129&$12&"8"7$32=& !! ]ZIhf&O&Z:\\\&-.129&$-TA"12=& !! ]I][&O&]\:\\\&-.129&$-TA"12=& !! xI]f&O&G#A9321&9FA3&=.S8&

GF"-$<.&N8"4.1%&;2278<:&Z[&H2D:&Z\]Z& Zg&

E9)%3K:9C/#H-,3%7:-#L7;,?)/#&()#M)*#N%&# O:9D:;,:#&:#P>,;*#%#O9,8%&)#F;:>*#A:9#Q,-%-',%;#%-*# F9)%78)#H44;,'%7:-/#

& F>/&:3)9R&012$%!.1/9&,8"%$7.8&VL@& 2-*>/&9=R&*83213$"8%283& S):.9%4(=R&@#28=$#2:&G$#"4.18"$:&$8=&J2=S..=&G"3E:&G$#"4.18"$& P>/,-)//#F(%;;)-.)R&*8$D#2&1$6"=&$8=&-.93R2b-"283&9-$#$D"#"3E&.4&-.%6A78<&$8=& 129.A1-29&4.1&oXG&'F"<FR6214.1%$8-2&-.%6A32)&S.1/#.$=9&$8=&.67%$#&A92&.4&$& %A#7R=$3$-28321&4..361"83+&*8F$8-2&3F2&12#"$D"#"3E&$8=&$>$"#$D"#"3E&.4&%"99".8R

• 1"7-$#&921>"-29+&

I:;>7:-R&

E)4;:=#M)*#N%&#T-&)949,/)#GMS#%/#&()#.9,*#,-A9%/&9>'&>9)#:A# E9)%3K:9C/#H-,3%7:-U/#JV@VVVW#':9)#9)-*)9#A%93#

026#.E&J2=&o$3&*832161"92&a"13A$#"r$7.8&$9&3F2&6#$~.1%&3.&1A8&%"99".8R-1"7-$#& 921>"-29&.8&J2=&o$3&*832161"92&C"8A_& &

GF"-$<.&N8"4.1%&;2278<:&Z[&H2D:&Z\]Z& Z[&

So, what’s not to like?

!! 93226&#2$18"8<&-A1>2z&9"<8"`-$83&"8>293%283&"8&FA%$8&2y.13+&

O! QF2&3F"8<9&3F$3&F$>2&D228&61"%$1"#E&9A--2994A#:&#"/2&@5&$8=&G.8=.1:&F$>2&31"2=&3.&$==1299&3F"9:&DA3&"3& "9&97##&8.8R31">"$#&

!! VA66.13&

O! QF"9&"9&.t28&$&-.8-218&4.1&-.%%21-"$#&A9219+& O! QF2&@#.DA9&X1.K2-3&"9&$&D293&2y.13&9A66.13&%.=2#+&&u.A&-$8&`#2&DA<9:&.1&928=&%$"#&3.&3F2&#"93:&DA3& 3F212&"9&8.&VC,&"8&12<$1=9&3.&SF$3&3F2&1296.892&7%2&c&TA$#"3E&S"##&D2+& O! QF212&"9&c&S$9&9.%2&-.%%21-"$#&9A66.13&$>$"#$D#2&

Y! N8">$&O&D$92=&"8&C"9#2:&93$132=&DE&9.%2&.4&3F2&.1"<"8$#&@#.DA9&62.6#2+&&?&S.A#=&=29-1"D2&3F2%&$9&'$%.8<&.3F21& 3F"8<9)&31E"8<&3.&D2&3F2&J2=o$3&.4&3F2&@1"=&S.1#=&'D2U21&6$-/$<"8<:&-.%%21-"$#&9A66.13:&23-++)& Y! ?(;&A92=&3.&61.>"=2&3F"9&$8=&A#7%$32#E:&"4&E.A&6$E&3F2%&%.82E&3F2E&S"##&92##&E.A&921>"-29:&DA3&?&=.8v3&/8.S& F.S&<..=&3F2E&S.A#=&D2&"8&3F$3&$128$&$8E&%.12+& Y! ?&$%&8.3&6.9"7>2:&DA3&6129A%$D#E&E.A&-$8&<23&9A66.13&4.1&G.8=.1&3F1.A<F&J2=o$3&9"8-2&"3&"9&3F2&F2$13&.4&3F2"1&

• #.A=&DA8=#2+&

!! j.V&

O! ?4&E.A&$12&8.3&="#"<283:&$&129.A1-2&-$8&<23&.>21SF2#%2=&DE&@1"=&S.1/#.$=9+&

!! 02DA<<"8<&

O! !"3F&921>"-29&$8=&K.D9&1A88"8<&#"U21$##E&$##&.>21&3F2&6#$823:&`<A1"8<&.A3&SF$3&$-3A$##E&S283&S1.8<&

• $8&D2&3.A<F+&

!! G.RV-F2=A#"8<&O&"4&E.A&822=&%A#76#2&129.A1-29&$3&%A#76#2&6#$-29&E.A1&61.D#2%&<239& 2_6.8287$##E&%.12&="b-A#3+&

GF"-$<.&N8"4.1%&;2278<:&Z[&H2D:&Z\]Z& Zx&

Questions?

Knowing the other and knowing oneself, In one hundred battles no danger. Not knowing the other and knowing oneself, One victory for one loss. Not knowing the other and not knowing oneself, In every battle certain defeat.

• Sun Tzu, The Art of War

Massively Parallel Computing and Argonne’s Storage System

It’s all about communication…

!! ,&31A2&93$32%283&"8&9.&%$8E&-.832_39&

O! (A9"8299&'E.A&61.D$D#E&/8.S&3F"9&D2U21&3F$8&?&=.)& O! ;$11"$<2&'%E&S"42&S.A#=&61.D$D#E&9$E&E.A&-213$"8#E&/8.S&3F"9&D2U21&3F$8&?&=.&"&)& O! ,8=&3.=$Ev9&3.6"-:&6$1$##2#&-.%6A78<&

!! !F$3&%$/29&$&9A621-.%6A321&9A621s&

O! QF2&"8321-.882-3:&"+2+:&3F2&-.%%A8"-$7.89&823S.1/&4.1&61.-299.1&3.&61.-299.1&

• .%%A8"-$7.8+&

Y! ?8&4$-3:&3F2&(#A2&@282cX&1A89&[f\;or&XXGhf\&-F"69:&DA3&3F2&823S.1/&"9&4$93&28.A<F&$8=&#.S& 28.A<F&#$328-E&3F$3&E.A&-$8&2b-"283#E&3F1.S&#.39&.4&3F2%&$3&3F2&61.D#2%+&

O! J2%2%D21&3F$3&3F2&="y2128-2&D23S228&@1"=&$8=&X$1$##2#&-.%6A78<&"9&"8=2628=283& 3$9/9&>9+&-..1="8$32=&3$9/9& O! G..1="8$7.8&12TA"129&-.%%A8"-$7.8&$8=&"4&E.A&$12&9628="8<&7%2&-.%%A8"-$78<& 'QF"8/&?c5&S$"3)&3F28&E.A&$128v3&-.%6A78<&

Y! !F"#2&"8&3F2.1E&E.A&-$8&.>21#$6&-.%%A8"-$7.8&$8=&-.%6A3$7.8:&"3&"9&.t28&3F2&8$3A12&.4&3F2& D2$93&3F$3&E.A&F$>2&3.&-.%6A32&9.%23F"8<:&928=&3F$3&129A#3&3.&E.A1&82"<FD.19:&S$"3&4.1&3F2%& 3.&1296.8=:&$8=&3F28&D$92=&.8&3F$3&1296.892:&-.%6A32&9.%2&%.12+&&u.A&F$>2&3.&S$"3+& Y! C$328-E&"9&$9&"%6.13$83:&"4&8.3&%.12&"%6.13$83&3F$8&D$8=S"=3F&

O! QF2&D"<<293&%$-F"829&A92&$&-A93.%&"8321-.882-3&'(#A2&@282:&G1$E&V2$V3$1&.1&@2%"8":& 0$S8"8<)+&&?4&3F2E&=.8v3&A92&$&-A93.%&"8321-.882-3:&?8`8"D$8=&"9&3F2&G5QV&"8321-.882-3& .4&-F."-2+&

GF"-$<.&N8"4.1%&;2278<:&Z[&H2D:&Z\]Z& eZ&

33

Blue Gene/P at ALCF

GF"-$<.&N8"4.1%&;2278<:&Z[&H2D:&Z\]Z&

34 34

ALCF Resources – Current

USER TEAMS

(via ESnet, UltraScienceNet, Internet2)

X!V#YP&

G=9,':3# IB,&'(#

01Z#YP& Z#OP&

?c5#

?c5#

G=9,':3#IB,&'(#F:34;)[#

Z#OP&

P r

• d

u c t i

• n

Eureka (Viz)

800 cores 50 NVIDIA S4 GPUs 100 TF

Intrepid

40 racks 160K cores 556 TF

Surveyor

1 rack 4K cores 13.9TF

00B&xff\9& 00B&xx\\9& 00B&xff\& V62-31$&C.<"-&Qxf\&

T e s t & D e v e l

• p

m e n t

S%*?::C/#\],?^## # J1#':9)/# # 1#_]2E2H#I!#SOL/&

Magellan

504 nodes 4032 cores

Retasked Oct, 2011

C l

• u

d

2-`-,$%-*# IB,&'(#

&VA%%21&Z\]\&

USER TEAMS

(via ESnet, UltraScienceNet, Internet2 now, ANI in 2011)

GF"-$<.&N8"4.1%&;2278<:&Z[&H2D:&Z\]Z&

Viz

3 TB RAM 1536 cores 48 K GPU cores

IB Federated Switch

Data Storage

240 GB/s write BW (aggregate) 28 PB capacity

BGQ Compute

208 TF 16 TB RAM 16 K cores

2P#

Test & Development WAN

100 Gb/s

PSab#c,-C/#\1[1SPa/^# 2P#bEM#\!SPa/^# T&()9-)&#\0VS$a/^#

Planned 2012 Delivery

2014 Tape Expansion

100 PB

BGQ Compute

10 PF 768 TB RAM 768 K cores

Mira

Login Nodes

2014 Storage Expansion

~478 GB/s write BW (aggregate for both phases) ~74 PB aggregate capacity both phases

Login Nodes

Data Storage

2.4 PB 20 GB/s write BW (aggregate)

T&()9-)&#

BGQ IO

8 PCI slots 128 GB RAM 128 cores

GridFTP Servers

12 servers

BGQ IO

384 PCI slots 6 TB RAM 6 K cores

Argonne Confidential

GF"-$<.&N8"4.1%&;2278<:&Z[&H2D:&Z\]Z& ef&

BlueGene/P Overview

36

850 MHz 8 MB EDRAM 4 processors 1 chip, 20 DRAMs 13.6 GF/s 2.0 GB DDR Supports 4-way SMP 32 Node Cards 1024 chips, 4096 procs 14 TF/s 2 TB 40 Racks 557 TF/s 80 TB RAM Cabled 8x8x16 Rack Intrepid System Compute Card Chip 435 GF/s 64 GB (32 chips 4x4x2) 32 compute, 0-2 IO cards Node Card Front End Node / Service Node System p Servers Linux SLES10 HPC SW: Compilers GPFS ESSL Cobalt

GF"-$<.&N8"4.1%&;2278<:&Z[&H2D:&Z\]Z&

Fabric Implementation

!! Used Myrinet 10g for the network

–! Over 1000 ports in use –! Ethernet from BG/P to Edge, MX elsewhere –! Uses CLOS / Fat tree topology with wormhole switching –! Pros

• ! 2z chip integrated standard Ethernet
• ! dumb switch, much cheaper (like 80%)
• ! Very high density (512 ports in 21U)
• ! low latency (not critical for us)

O! G.89&

Y! a21E&F"<F&=289"3E&"z&F$1=&3.&S.1/&S"3F& Y! (#$-/&D._z&%$8$<2%283&3..#9&S2$/z&=2DA<<"8<& ="b-A#3& Y! B22=2=&-#.92&-..1="8$7.8&S"3F&>28=.1&3.&$-F"2>2& '$<<1299">2)&6214.1%$8-2&<.$#9& Y! QE6"-$#&82S&61.=A-3&"99A29& H"#2&V21>219 & (#A2&@282cX &

Core Switches Edge Switches Edge Switches

GF"-$<.&N8"4.1%&;2278<:&Z[&H2D:&Z\]Z& eg&

Disk Storage

!! Key Drivers

–! Performance is paramount –! Large files, streaming access is the norm

!! Use of SATA for cost / capacity, but still needed performance !! Corruption issues a significant concern. !! Configured a storage cell

–! (2) storage arrays with a file server / switch rack in between –! Developed models that could ripple changes through the design (impact on switches, ports, power, space!)

GF"-$<.&N8"4.1%&;2278<:&Z[&H2D:&Z\]Z& e[&

Tape is dead! Long Live Tape

!! The reports of my death are greatly exaggerated! at least for our application

!! Space

–! Both of these pictures show approximately 8PB of high density storage

!! Power

–! (2) L6-30P plugs < 1KW in normal operation –! (128) L6-30P plugs approx 250KW in normal operations

GF"-$<.&N8"4.1%&;2278<:&Z[&H2D:&Z\]Z& ex&

Scientists are Pack Rats!

!! Key Drivers

–! Large, unique, relatively uncompressible data –! Data loss is undesirable, but not catastrophic –! Restore time is not critical –! Once archived, it is rarely accessed –! We do our own maintenance

!! Three primary tasks

O! N921&"8"7$32=&$1-F">2I&&N929&oXVV+&&;.>29&$12&A921&"8"7$32=&61"%$1"#E&>"$&3F2&oV?&-.%%$8=:& DA3&9.%2&A9219&A92&oQ,J+&&!2&"8328=2=&3.&%$/2&oQ,J&3F2&61"%$1E&-#"283:&DA3&"3&F$9&`#2&9"r2& $8=&6$3F&#28<3F&#"%"39&3F$3&S2&F"3&12<A#$1#E+&&QF"9&"9&3F2&DA#/&.4&3F2&>.#A%2+&&V-"287939&$12& 6$-/&1$39+&&QF2E&82>21&3F1.S&$8E3F"8<&$S$E+&&V.&SF28&3F2E&$12&=.82&S"3F&3F2&=$3$:&3F2E& 6A3&"3&.8&3$62&"8&-$92&3F2E&=2-"=2&3F2E&S$83&3.&12R$8$#Er2&"3&Z\&E2$19&41.%&8.S+& O! o.%2&D$-/A69I&&H"193&#"82&.4&=242892&"9&@XHV&98$69F.39+&&Q$62&A929&oXVV+&&QF292&$12&"8"7$32=& DE&A9&$9&1..3+&&&G#"283&93.1E&"9&3F2&9$%2&$9&$D.>2:&%.93#E&oV?:&S2&S"9F&S2&-.A#=&A92&oQ,J+&& QF"9&"9&3F2&DA#/&.4&3F2&`#2&-.A83&'Ä#.39Ä&.4&9%$##&`#29)+&&!2&297%$32&"3&S.A#=&3$/2&A9&i&S22/9& 3.&=.&$&-.%6#232&1293.12&.4&$&-$3$931.6F"-&4$"#A12&.4&.A1&F.%2&`#2&9E932%&IR'+& O! G1"7-$#&9E932%&D$-/A69I&&N929&,%$8=$+&&QF292&$12&4.1&.A1&"83218$#&A92&.8#E+&&B"<F3#E& "8-12%283$#&D$-/A69&1A8&]f&F.A19+&&&&;.83F#E&4A##&"9&$D.A3&hQ(&$8=&1A89&$D.A3&h&=$E9+&&!2& 6#$8&3.&S1"32&=A$#&-.6"29:&.82&3.&2$-F&#"D1$1E+&&!2&F$=&"8328=2=&3.&D2&=."8<&9.&$#12$=E:&DA3& 823S.1/"8<&D23S228&3F2&=$3$&-283219&612>2832=&"3+&&!2&9F.A#=&F$>2&3F$3&129.#>2=&3F"9&E2$1+&& H.1&8.S:&S2&.y9"32&3F2&3$629&"8&.A1&.b-29+&

GF"-$<.&N8"4.1%&;2278<:&Z[&H2D:&Z\]Z& h\&

How does tape fit in my world?

!! Q$62&"9&-A11283#E&8.3&%"99".8&-1"7-$#&3.&A9+&&?4&S2&#.92&$--299&3.&$&#"D1$1E:&S2&12-.1=&"3:& DA3&"3&-$A929&8.&9"<8"`-$83&"99A29+&&58&3F2&.3F21&F$8=:&9-F2=A#2=&$>$"#$D"#"3E&4.1&oXVV& S$9&xx+gÅ+&&B.32&3F$3&.A1&4$-"#"3E&3$/29&$&%$"8328$8-2&.A3$<2&2>21E&.3F21&;.8=$E+& !! 5A1&=$3$&A92&F$9&D228&8.&SF212&82$1&SF$3&S2&61.K2-32=+&&?&D2#"2>2&"3&"9&D2-$A92&S2& =.8^3&1A8&$&9-1$3-F&`#2&9E932%&#"/2&.3F21&4$-"#"729+&&!2&#23&.A1&A9219&/226&3F212&=$3$& .8&="9/&;NGo&#.8<21&3F$8&.3F21&4$-"#"729+&&582&6$17-A#$1&A921&F$9&F$=&82$1#E&\+fX(& .8&="9/&4.1&Z&E2$19+& !! QF.A<F&8.3&"8"7$##E&6#$882=:&S2&3../&$=>$83$<2&.4&3F2&#"D1$1E&>"13A$#"r$7.8&42$3A129+&& !2&F.93&3S.&.3F21&=">"9".89&.8&.A1&#"D1$1"29+&&V$>2=&3F2&6A1-F$92&.4&9%$##&#"D1$1"29& $8=&3F2&$99.-"$32=&$=%"8"931$7.8&$8=&%$"8328$8-2&-.939+& !! GF$8<29&-.%"8<I&

O! !2&F.62&3.&A92&@o?:&SF"-F&A929&oXVV&$9&$&oV;&4.1&@XHV+&

Y! QF"9&S"##&%2$8&3$62&A9$<2&S"##&<.&A6&=1$%$7-$##E+& Y! QF"9&S"##&$#9.&%2$8&3F$3&3$62&$--299&S"##&D2-.%2&%A-F&%.12&-1"7-$#+&

O! 921>21&D$-/A6&',%$8=$)&#.$=&S"##&=1.6+&&B2S&921>219&S"##&%.A83&=1">29&41.%&.A1&B23,66& $8=&S2&S"##&A92&98$69F.39&$8=&12=A-2&3F2&D$-/A6&"8321>$#+&

!! ;E&61.D#2%9&c&-.8-2189&

O! Ze\&%"##".8&9%$##&`#29+&&QF2&%23$=$3$&/"##9&oXVV+&&QF2&6214.1%$8-2&.8&9%$##&`#29&"9&6..1& '297%$32=&i&S22/9&3.&1293.12&F.%2)& O! 0$3$&?832<1"3EI&?9&3F2&=$3$&97##&<..=z&126#"-$7.8&>9+&J,?Q:&23-++&

GF"-$<.&N8"4.1%&;2278<:&Z[&H2D:&Z\]Z& h]&

Tape Server (2) LTO4 x10 for Archive

Tape Block Diagram

G=9,':3#IB,&'(#F:34;)[#

&'()*!(+,-

Z#OP#a#XVW#SPa/# H9'(,8)#5-;=&

."/0-

X!V#YP#ZW#SPa/# dVe#H9'(,8)#a#0Ve#P%'C>4&

1)%2(*#-'34!0/-5*($674-

c:'%;#I)98)9#E,/C#f#0YP# Y=4,'%;#P%'C>4&

Tape Server

• !Quad core, 4GB RAM
• !(1) 10Gbs Myricom
• !(2) 4 Gbs FC HBA

(2) LTO4 x2 for Backup

• !Quad core, 4GB RAM
• !(1) 10Gbs Myricom
• !(2) 4 Gbs FC HBA

Disk Server

• !Quad core, 4GB RAM
• !(1) 10Gbs Myricom
• !(1) IB 4x DDR HCA

x8 for Both

Spectralogic 8 frame T950 10,000 slots (24) LTO4 tape drives DDN 9900 (480) 1TB drives

GF"-$<.&N8"4.1%&;2278<:&Z[&H2D:&Z\]Z& hZ&

Tape Software

!! Using HPSS (www.hpss-collaboration.org)

–! large files are a key driver for us –! Implements tape striping (think RAID 0)

• ! Allows reasonable access times to TB sized files
• ! Unexpected benefit when paired with Spectralogic: Single tape pick can get all the tapes

–! Actually need a feature added to realize this because HPSS puts the tape back in nearest open slot not where it came from

–! Single name space

• ! Hides multiple libraries
• ! Uses DB2; Can scale to billions of filenames

–! Weak on the client side.

!! Using Amanda for backup

–! Our backup needs are small, but! –! There are significant limitations to Amanda and we are investigating commercial solutions (TSM, NetBackup, etc)

GF"-$<.&N8"4.1%&;2278<:&Z[&H2D:&Z\]Z& he&

A recent snapshot of our HPSS system

GF"-$<.&N8"4.1%&;2278<:&Z[&H2D:&Z\]Z& hh&

Impact of Multiple Cores

!! HPC requires IO nodes !! Over the next 5 years you will see 16, 32, 64, maybe even more cores !! Two possibilities

–! Make the storage dumber

• ! Bring the storage closer and move functionality out of HW and into

SW

• ! Software is the key

–! Reliability, manageability, Metadata

• ! Google File System is a variant of this

–! Make the storage smarter

• ! Put the IO node(s) in the storage
• ! Ensure good match between IO node and storage
• ! Needs to be flexible on amount of compute

–! Ratio of IO to storage varies widely

GF"-$<.&N8"4.1%&;2278<:&Z[&H2D:&Z\]Z& hf&

Data Integrity

!! RAIT: Nothing new, but some interesting possibilities!

–! Stripe over n tapes, but need only m to restore –! Write n+1 at first, asynchronously add extra redundancy –! If drives are busy, load n or n+1 and compute the rest

• ! Note the multi-core makes this even more attractive

!! Cleversafe

–! DISCLAIMER: I am on the technical advisory board and I have stock in this company. –! Arbitrary level of parity protection: Split the data across 57 (potentially geographically distributed) devices and as long as I have 39 I can rebuild it

• ! The numbers 57 and 39 are just examples. Can trade storage space for reliability
• ! Undetected data integrity faults can mathematically be in the hundreds of millions of

years.

• ! Their focus is reliable and cheap, but they are not fast! sounds kind of like tape!

–! Still random access, high reliability, but requires power. –! No file system, object based storage system

• ! Possible replacement for tape?

–! Particularly if the restore time from tape is not acceptable (photo sites) –! Shutterfly uses this for their photo storage

GF"-$<.&N8"4.1%&;2278<:&Z[&H2D:&Z\]Z& hi&

,CGH&5621$7.8$#&,99299%283&O&,A<A93&ZfRZi:&Z\]]&

hg&

Laser-Plasma Interactions for the National Ignition Facility

Denise Hinkel, LLNL

Science and Approach Key Impact ALCF Contributions

• ! N9"8<&-.=2&6He0&9"%A#$78<&

D.3F&1$="$7.8&$8=& FE=1.=E8$%"-9&

• ! ,&#$1<2R9-$#2&9"%A#$7.8&.4&

#$921R6#$9%$&"8321$-7.8&"8& 3F2&B$7.8$#&?<8"7.8&H$-"#"3E& 'B?H)&#../"8<&$3&8.8RA8"4.1%&

• 1.99RD2$%&2821<E&31$89421&
• ! 0.AD#2&TA$=RD2$%&

9"%A#$7.89&F$>2&9F.S8&$8& "8-12$92&.4&12p2-7>"3E&.A3&.4& 3F2&3$1<23&

• ! *_6$8="8<&3.&%.12&D2$%9&"9&
• 1"7-$#&3.&"%61.>"8<&3$1<23&

=29"<8&

ONgI2FI#5Q#OcHIGHI#0Z@# VhXJ01#\1V00^#

&

• ! ?%6$-32=&=29"<8&.4&3$1<23&4.1&

4A3A12&B$7.8$#&?<8"7.8&H$-"#"3E& 2_621"%2839+&

• ! HA9".8&2821<E&61.%"929&3.&

D1"8<&9A93$"8$D#2:&<1228& 2821<E&61.=A-7.8&3.&3F2& S.1#=&&

• ! 02DA<<"8<&$&42S&-1"7-$#&"99A29&
• ! G..1="8$32=&#$1<2&-$%6$"<89&

VJV&J2p2-32=& C"<F3&

V"8<#2RTA$=&D2$%& 61.6$<$7.8& 0.AD#2RTA$=&D2$%& 61.6$<$7.8& 8918:;-- <=>-

,CGH&5621$7.8$#&,99299%283&O&,A<A93&ZfRZi:&Z\]]&

h[&

Lattice QCD Paul Mackenzie, Fermilab

Science and Approach Key Impact ALCF Contributions

• ! X1.>"=2&-1A-"$#&F"<F&

612-"9".8&#$M-2&jG0&

• $#-A#$7.89&822=2=&4.1&82S&

.1&"8R61.<1299&2_621"%2839& $8=&$8$#Er2&129A#39&41.%& 2_621"%2839&&

• ! N92=&"8&%E&F"<F&2821<E&$8=&

8A-#2$1&6FE9"-9&

C$M-2&jG0&F$9&=2#">212=& 299287$#&129A#39&3.& 2_621"%283$#&61.<1$%9&

• ! Jo?G&R&o2#62=&3.&`1%#E&
• .8931$"8&F2$>ER".8&-.##"9".8&

%.=2#9&4.1&3F2&`193&7%2+&& H"193&$8=&92-.8=&%.93&-"32=& 6$6219&"8&CjG0&9"8-2&Z\\i&

• ! H21%"#$D&O&G$#-A#$7.89&.4&

NVjG0&%2%D219&-.%D"82=& S"3F&2_621"%283$#&129A#39& F$>2&$##.S2=&%$8E&.4&3F2& 4A8=$%283$#&6$1$%23219&.4& 3F2&V3$8=$1=&;.=2#&3.&D2& =2321%"82=&%.12&$--A1$32#E& 3F$8&2>21&D24.12+& 0"9-.>212=&$&82S:&%A-F&%.12& 2b-"283:&S$E&3.&-$#-A#$32&3F2&

• ?Vj&l421%".8&4.1-2m+&&&
• ! J2=A-29&HC5XV&]\_&&
• ! ?%61.>29&6$1$##2#&2b-"28-E&
• ! J261292839&]\Rf\Å&.4&3.3$#&

7%2& & V"%"#$1&"%61.>2%2839&S212&%$=2& 4.1&3F2&<$A<2&4.1-2&1.A7829+&&&

• ! Ç]\Å&.4&1A87%2&$8=&"9&8.S&

ZRe_&4$9321&"8&3F2&;?CG&-.=2& & GF$8<29&$12&-$63A12=&"8&3F2&CjG0& V-"0,G&#"D1$1"29+& 8918:;-- <=>-

Laiho, Lunghi, and Van de Water

,CGH&5621$7.8$#&,99299%283&O&,A<A93&ZfRZi:&Z\]]&

hx&

Deflagration-to-Detonation Transition in Reactive Gases

Alexei Khoklov, University of Chicago

Science and Approach Key Impact ALCF Contributions

• ! H"193&61"8-"6#2:&12$-7>2&p.S&

="12-3&8A%21"-$#&9"%A#$7.89&

• ! g&%"-1.8&129.#A7.8&9F.-/&

3AD2&"8&G5Z&S"3F&F2$3&

• .8=A-7.8&$8=&"9.3F21%$#&

S$##9&F$9&2_-2##283&$<122%283& S"3F&2_621"%283&

• ! ;.12&-.%6#"-$32=&6FE9"-9&4.1&

82_3&9A"32&.4&9"%A#$7.89&

• ! 029"<8&9$42&9E932%9&4.1&4A3A12&

FE=1.<28&2821<E&9E932%9&DE& <$"8"8<&"89"<F3&"83.&3F2& 6FE9"-$#&%2-F$8"9%9&.4&3F2& DA18"8<&$8=&=23.8$7.8&.4& FE=1.<28R._E<28&%"_3A129&

• ! J2>2$#2=&$&931.8<&#.$=&"%D$#$8-2&

$8=&61292832=&$&12-.%%28=2=& 9.#A7.8+&&G.##$D.1$78<&.8& "%6#2%283$7.8+&

• ! a"9A$#"r$7.8&32$%&F2#62=&

>"9A$#"r2&=$3$&"8&a?V?Q&

• ! Q.&6126$12&4.1&"8"7$#&?BG?Q*&

$S$1=:&,CGHI&?%6#2%2832=& 5628;X&$##.S"8<&A92&.4&$##&-.129& .8&3F2&8.=2&<">"8<&e_&9622=A6& $8=&12S1.32&?c5+& &

8918:;-- ?@>-

Q2%621$3A12&"8&$&3F122R ="%289".8$#&B$>"21RV3./29& `193R61"8-"6#29&="12-3& 8A%21"-$#&9"%A#$7.8&.4&$& ;$-FÉe&12p2-32=&9F.-/& D"4A1-$7.8&"8&$&FE=1.<28R ._E<28&%"_3A12&"8&$&9TA$12&

• F$882#+&&&

,CGH&5621$7.8$#&,99299%283&O&,A<A93&ZfRZi:&Z\]]&

f\&

High Reynolds Number Turbulent Boundary Layers

Robert Moser, University of Texas

Science and Approach Key Impact ALCF Contributions

• ! o"<F293&2>21&J2E8.#=9&

8A%D21&9"%A#$7.89&4.1& 96$7$##E&2>.#>"8<& "8-.%61299"D#2&3A1DA#283& D.A8=$1E&#$E219&

• ! 0"12-3&8A%21"-$#&9"%A#$7.8&

'0BV)&.4&F"<F&J2E8.#=9& 8A%D21&3A1DA#283&p.S9&"8&$& D.A8=$1E&#$E21&

• ! @$3F21"8<&3F2&12TA"12=&

3A1DA#28-2&93$797-9&"8& $8$#E9"9&12TA"129&9"<8"`-$83&

• .%6A3$7.8&
• ! (.A8=$1E&#$E219&$12&-2831$#&3.&

3F2&2821<E&#.9929&"8& 31$896.13$7.8&

• ! N8=2193$8="8<&3F292&p.S9&-$8&

"%61.>2&3F2&=29"<8&$8=& 2b-"28-E&.4&>2F"-#29+&&QF292& =29"<89&$12&"%62=2=&DE&3F2& #$-/&.4&$--A1$32&%.=2#9&.4&3F2& 3A1DA#283&6F28.%28$&

• ! ?8"7$#&6.13&$8=&6214.1%$8-2&

.67%"r$7.89&4.1&?83126"=&SF"-F& =1.>2&4A13F21&.67%"r$7.89&DE& ;.921v9&32$%+&

• ! Q2978<&3F12$="8<&6214.1%$8-2&

4.1&;"1$&

' ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! !

!

! ! ! ! ! ! ! ! ! ! ! ! ! !

! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! !! QF"9&9F.S9&3F2&3A1DA#283c8.8R3A1DA#283&"83214$-2&$3&3F2&D.A8=$1E&#$E21z&12=&$12&3A1DA#283&r.829& $8=&D#A2&$12&8.8R3A1DA#283&r.829+&&J.3$7.8$#&p.S&"9&$&/2E&="y2128-2&D23S228&D.A8=$1E&#$E21&p.S& $8=&"83218$#&p.S9&'"8&-F$882#9&$8=&6"629)+&&N8=2193$8="8<&3F2&1.3$7.8$#&p.S&S"##&&12>2$#&%.12& "84.1%$7.8&$D.A3&F2&931A-3A12&.4&3F2&p.S&9F.S8&F212+&

8918:;-- A=>-

,CGH&5621$7.8$#&,99299%283&O&,A<A93&ZfRZi:&Z\]]&

Reducing Toxic Gas through Metal Catalysis

Jeff Greeley, Argonne National Laboratory

f]&

Science and Approach Key Impact ALCF Contributions

• ! 02321%"8"8<&3F2&9"r2&.4&

8$8.6$17-#29&3.&A92&4.1&

• $3$#E939&4.1&.67%$#&129A#39&"9&

="b-A#3&S"3F&6FE9"-$#& 2_621"%2839&

• ! X#$78A%&8$8.6$17-#29&
• .8>21<2&%A-F&TA"-/21&3F$8&

<.#=&6$17-#29&']hg&$3.%9&>9& fi]&$3.%9)&

• ! 5'Be):&12$#&96$-2:&<1"=RD$92=&

0289"3E&HA8-7.8$#&QF2.1E& '0HQ)&-.=2:&@X,!:&4.1&.A1& 8$8.-$3$#E7-&%.=2#"8<& 2y.139+&&

• ! *#2-31.-$3$#E9"9&4.1&4A2#&-2##9&
• ! ;23$#&-$3$#E9"9&.4&-$1D.8&

%.8._"=2&3.&-$1D.8&="._"=2&

• ! *8$D#2&=29"<8&.4&"%61.>2=&
• $3$#E7-&9E932%9&'#"/2&
• $3$#E7-&-.8>213219)&4.1&$&

1$8<2&.4&"8=A931"$#&A929&

• ! J2=A-2=&%2%.1E&4..361"83&

$8=&$==2=&#$E21&.4& 6$1$##2#"r$7.8&

• ! V-$#2=&41.%&f]Z&-.129&3.&

]e]:\gZ&-.129&

• ! X$1$##2#&?c5&"%6#2%283$7.8&

1A88"8<&h\Å&.4&62$/&$3&eZL&

• .129&4$-"#"3$78<&2$9"21&A92&.4&

?83126"=&

• ! ?=287`2=&>21E&="b-A#3&

V-$C,X,GL&%2%.1E&DA<& 8918:;- ?B>-

QF2&-F$1<2&=289"3E& ="y2128-2&.4&$&-$1D.8& %.8._"=2&%.#2-A#2&O& .82&._E<28&$3.%&$8=& .82&-$1D.8&$3.%&RR& $=9.1D2=&'$=F212=&3.& 3F2&9A14$-2)&.8&$&<.#=& 8$8.6$17-#2&.4&e\x& $3.%9&&

,CGH&5621$7.8$#&,99299%283&O&,A<A93&ZfRZi:&Z\]]&

fZ&

Multiscale Blood Flow Simulations

George E. Karniadakis, Brown University

Science and Approach Key Impact ALCF Contributions

• ! V3A=E&$;::*#i:B#"8&$&6$7283R

962-"`-&-212D1$#&>$9-A#$3A12& S"3F&$&$9%,-#%-)>9=/3&$8=& $8$#Er2&3F2&"8321$-7.8&.4& D#..=&-2##9&S"3F&3F2&$1321"$#& S$##9&3F$3&#2$=&3.&&(9:3$>/# A:93%7:-#$8=&2>283A$#& $82A1E9%&1A63A12+&

• ! ;$66"8<&2_$-3#E&F.S&9)*#

$;::*#');;/#%.>2&3F1.A<F&3F2& D1$"8&$8=&93A=E&VV&$8=&G;& ="92$929+&

• ! N9"8<&':34>&)9/:&/',)-7/&/#

$8=&'%9*,:;:.,/&/#S"##&-.8=A-3& >"13A$#&2_621"%2839&3.&93A=E&

• 212D1$#&D#..=&p.S&$8=&

"%61.>2&*,%.-:/,/#$8=& &9)%&3)-&+&

• ! F)9)$9%;#%-)>9=/3/#.--A1&"8&

A6&3.&he&.4&3F2&<2821$#& 6.6A#$7.8:&#2$="8<&3.&931./29& "8&.>21&!V@VVV&,%21"-$89&2$-F& E2$1+&

• ! I,'C;)#');;&%-)3,%#'$&-F1.8"-&

"8p$%%$3.1E&="92$92)&$y2-39& j1@VVV#"8=">"=A$#9&"8&NV,+&

&

&

• ! F)9)$9%;#3%;%9,%&"9&$&12=&D#..=&
• 2##&="92$92:&S"3F&9"<8"`-$83&

%.13$#"3E&$8=&%.93#E&$y2-39&

• F"#=128+&
• ! X214.1%$8-2&3A8"8<&$8=&

.67%"r$7.8I&&

• !"%61.>2=&%299$<"8<&1$32&
• !"%61.>2=&"8931A-7.8&9-F2=A#"8<&
• !12=A-2=&%2%.1E&$--29929&
• !?%61.>2=&=$3$&$#"<8%283&
• ! V-$#2=&A6&3.&0J0@Vj1#':9)/#\PSaO^+&
• ! @.1=.8&(2##&`8$#"93&VG&Z\]]+&

&

,S$1=&3.&12-.<8"r2&.A393$8="8<& $-F"2>2%283&"8&oXG&$66#"-$7.89##

5%("C(*D0-*CE-4%($#0- 4+*70-"F-!+0-)0E- G#""E-(0##4-HI51J-#

8918:;- <=>--

,.&(304%!#%5#67$8#.!#0#/*88*3#

,CGH&5621$7.8$#&,99299%283&O&,A<A93&ZfRZi:&Z\]]&

fe&

Materials Design from First Principles

Larry A. Curtiss, Argonne National Laboratory

Science and Approach Key Impact ALCF Contributions

J2=A-2&3F2&9"<8"`-$83&

• F#.1"8$32=&S$932&41.%&

61.=A-"8<&61.6E#282&

• ! VAD8$8.%2321&,<&-#A93219&

%$E&61.>"=2&6$3FS$E9&3.& .>21-.%2&3F"9&S$932& 61.D#2%+&

• ! V31A-3A12&.4&/2E&%.#2-A#2&

.67%"r2=&$8=&93A=E&

• .%6#232=&.8&6$13&.4&3F2&

61.-299& V3A=E&3F2&31$894.1%$7.8&.4& D".4A2#9&S"3F&2b-"283&-$3$#E939&

• ! V31A-3A12&.4&Ñ15Z&8$8.D.S#9&

"8&=1E&$8=&FE=1._E#$32=& $#A%"8$&F$>2&D228&.67%"r2=&

• !

?8&9A66.13&.4&3F2&?8973A32&4.1& ,3.%"-R2b-"283&GF2%"-$#& Q1$894.1%$7.8&'?,GQ):&$8&*821<E& H1.87219&J292$1-F&G28321&'*HJG)&

• !

P+&XFE9+&GF2%+&G:&Z\]]:&]]f&'hh):&66& Z]g[fOZ]gx\&

• ! X1.>"=2&3F2&4A8=$%283$#&

A8=2193$8="8<&$8=&612="-7.89& 822=2=&3.&=29"<8&82S& %$321"$#9&4.1&-$3$#E9"9&$8=& 2821<E&93.1$<2&

• ! V"<8"`-$83&7%2&S$9&96283&

%283.1"8<&h&6.93R=.-9&.8&A9"8<& @X,!&$8=&F.S&3.&A92&?83126"=&

• ! (A"#="8<&.y&.4&9"<8"`-$83&2y.13&

.8&3F2&6214.1%$8-2&$8=& 9-$#$D"#"3E&.4&@X,!&

QF"13ER3F122&$3.%&9"#>21&-#A9321&3F$3& "9&D2"8<&93A="2=&$9&$&82S&-$3$#E93&4.1& 61.6E#282&26._"=$7.8+&&&& B$8.D.S#&"8&Ñ15Z&9A14$-2&']8%& ="$%+)&3F$3&"9&D2"8<&93A="2=&4.1& D".%$99&-.8>219".8&3.&4A2#9&&&&&&&

KL11- M=>-

Faster Design of Better Jet Engines

!! V$>2&-.93&$8=&7%2&DE&=29"<8"8<&28<"829& 3F1.A<F&9"%A#$7.8&1$3F21&3F$8&DA"#="8<& %.=2#9& !! Q2-F8.#.<"29&41.%&9"%A#$7.89&8.S&D2"8<& $66#"2=&3.&82_3&<2821$7.8&F"<FR2b-"28-E&& #.SR2%"99".8&28<"829& !! ,&/2E&28$D#21&4.1&3F2&=263F&.4& A8=2193$8="8<&822=2=&3.&822=2=&3.&%223& 2%"99".89&<.$#9&

!! ?c5&$#<.1"3F%&12=29"<8&9622=9&A6& 9"%A#$7.89&DE&e_&&

54

Peter Bradly Pratt&Whitney

Science Challenges

?BG?Q*&

GF"-$<.&N8"4.1%&;2278<:&Z[&H2D:&Z\]Z&

Insight into Parkinson’s Disease

!! X$1/"89.8v9&0"92$92&$y2-39&f& %"##".8&"8&NV&$8=&*A1.62& !! ?8-12$92=&$<<12<$7.8&.4&!9 8:!(-3*.!&61.32"8&"9&3F.A<F3&3.&#2$=& 3.&F$1%4A#&6.12R#"/2&931A-3A129&"8& FA%$8&%2%D1$829& !! V"%A#$7.89&9F.S&!R9E8A-#2"8&

• .%6#2_29:&$8=&"R9E8A-#2"8&

612>2839&-12$7.8&.4&61.6$<$78<& !R9E8A-#2"8&-.%6#2_29&

ff&

Science Methods and Challenges

Igor Tsingelny University of California, San Diego

!! N9"8<&B,;0&$8=&;,X,V&.8&(#A2& @282&$3&,CGH&$8=&V0VG& !! !R9E8&61.32"89&$12&A8931A-3A12=&

?BG?Q*&

!-synuclein coformer at membrane (left), completed pentamer (above), evolution of !synuclein

• ver 5ns (right)

GF"-$<.&N8"4.1%&;2278<:&Z[&H2D:&Z\]Z&

Questions?

Knowing the other and knowing oneself, In one hundred battles no danger. Not knowing the other and knowing oneself, One victory for one loss. Not knowing the other and not knowing oneself, In every battle certain defeat.

• Sun Tzu, The Art of War