85
4th RICC workshop ご案内 第15回さくらの夕べ in 札幌 An invitation for 4th RICC workshop 柏崎 礼生 Hiroki Kashiwazaki

4th RICC workshopのご案内

Embed Size (px)

DESCRIPTION

第15回さくらの夕べin札幌のLT内容です。ちょっと補っております。

Citation preview

Page 1: 4th RICC workshopのご案内

4th RICC workshopの ご案内

第15回さくらの夕べ in 札幌

An invitation for 4th RICC workshop

柏崎 礼生 Hiroki Kashiwazaki

Page 2: 4th RICC workshopのご案内

小樽市桂岡町から きました

Page 3: 4th RICC workshopのご案内
Page 4: 4th RICC workshopのご案内

4th RICC workshop @Okinawa 2014/3/27(Thu)~28(Fri)

Page 5: 4th RICC workshopのご案内

RICC

Page 6: 4th RICC workshopのご案内

俵屋宗達: 風神雷神図 (1624ころ?) Soutatsu Tawaraya: Fujin Raijin-zu

Page 7: 4th RICC workshopのご案内

RICC

RICC

地域間 インタークラウド

分科会

Page 8: 4th RICC workshopのご案内

雲内放電 Inter Cloud Lightening

Page 9: 4th RICC workshopのご案内

4th RICC workshopの ご案内

第15回さくらの夕べ in 札幌

An invitation for 4th RICC workshop

柏崎 礼生 Hiroki Kashiwazaki

4th RICC workshop @Okinawa 2014/3/27(Thu)~28(Fri) RICC

俵屋宗達: 風神雷神図 (1624ころ?) Soutatsu Tawaraya: Fujin Raijin-zu

RICC

RICC

地域間 インタークラウド

分科会雲内放電 Inter Cloud Lightening 82p/5min TOYAMA site

OSAKA site

TOKYO site before Migration

Copy to DR-sites

Copy to DR-sites

live migration of VM between distributed areas

real time and active-active features seem to be just a simple "shared storage". Live migration is also possible between DR sites

(it requires common subnet and fat pipe for memory copy, of course)

after Migration

Copy to DR-sites 広域分散 仮想化環境

Distcloud DR Disaster Recovery

1978Sun Information Systems

mainframe hot site ‘80-’90 Realtime

Processing POS point of sales

’90-’00 the Internet 2001.9.11 September 11 attacks

2003.8.14 Northeast blackout of 2003 in Japan 2011.3.11

The aftermath of the 2011 Tohoku earthquake and tsunami BCP

Business Continuity Plan

群馬 Gunmma prefecture

石狩 Ishikari city

2つで 十分ですよ?

国立情報学研究所

Kitami Institute of Technology

University of the Ryukyus

北見工大

琉球大学

SINET 最長

XenServer 6.0.2

CloudStack 4.0.0

XenServer 6.0.2

CloudStack 4.0.0

problems

shared storage ≒50ms

RTT > 100ms Distributed

Storage

requirement 64 256 1024 4096 16384 65536 262144 1.04858e+06 4.1943e+06 1.67772e+07 6.71089e+07 4

16 64

256 1024

4096 16384

0

20000

40000

60000

80000

100000

120000

Kbyt

es/s

ec

File size in 2^n KBytes

Record size in 2^n Kbytes

0

20000

40000

60000

80000

100000

120000

High Random R/W Performance

POSIX準拠 interface protocl

NFS, CIFS, iSCSI

Global VM migration is also available by sharing "storage space" by VM host machines. Real time availability makes it possible. Actual data copy follows.

(VM operator need virtually common Ethernet segment and fat pipe for memory copy)

TOYAMA site

OSAKA site

TOKYO site before Migration

Copy to DR-sites

Copy to DR-sites

live migration of VM between distributed areas

real time and active-active features seem to be just a simple "shared storage". Live migration is also possible between DR sites

(it requires common subnet and fat pipe for memory copy, of course)

after Migration

Copy to DR-sites

Fileblock block block

block block block

block block block

Meta Data

consistent hash

backend (core servers) NFS

CIFS iSCSI

redundancy = 3

r = 2ACK

r = 1

r = 0

write

redundancy = 3

ACK

r = 2 e = 0

r = 1 e = 0

r = 0 e = 1

r = -1 e = 2

external

Hypervisor

VM 金沢大学 広島大学

iozone -aceI

write

64 256 1024 4096 16384 65536 262144 1.04858e+06 4.1943e+06 1.67772e+07 6.71089e+07 4 16

64 256

1024 4096

16384

0

20000

40000

60000

80000

100000

120000

Kbyt

es/s

ec

File size in 2^n KBytes

Record size in 2^n Kbytes

0

20000

40000

60000

80000

100000

120000

64 256 1024 4096 16384 65536 262144 1.04858e+06 4.1943e+06 1.67772e+07 6.71089e+07 4

16

64

256

1024

4096

16384

File size in 2^n KBytes

Record size in 2^n Kbytes

0

20

40

60

80

100

120

10MB 100MB 1GB 10GB

Thro

ughp

ut (M

B/s

ec)

File size

write rewrite read reread

random read random write bkwd read

stride read fwrite freadlegend

record rewrite

0

20

40

60

80

100

120

10MB 100MB 1GB 10GB

Thro

ughp

ut (M

B/s

ec)

File size

0

20

40

60

80

100

120

10MB 100MB 1GB 10GB

Thro

ughp

ut (M

B/s

ec)

File size

0

20

40

60

80

100

120

10MB 100MB 1GB 10GB

Thro

ughp

ut (M

B/s

ec)

File size

0

20

40

60

80

100

120

10MB 100MB 1GB 10GB

Thro

ughp

ut (M

B/s

ec)

File size

0

20

40

60

80

100

120

10MB 100MB 1GB 10GB

Thro

ughp

ut (M

B/s

ec)

File size

0

20

40

60

80

100

120

10MB 100MB 1GB 10GB

Thro

ughp

ut (M

B/s

ec)

File size

0

20

40

60

80

100

120

10MB 100MB 1GB 10GB

Thro

ughp

ut (M

B/s

ec)

File size

0

20

40

60

80

100

120

10MB 100MB 1GB 10GB

Thro

ughp

ut (M

B/s

ec)

File size

0

20

40

60

80

100

120

10MB 100MB 1GB 10GB

Thro

ughp

ut (M

B/s

ec)

File size

0

20

40

60

80

100

120

10MB 100MB 1GB 10GB

Thro

ughp

ut (M

B/s

ec)

File size

従来方式 Exage/Storage

広域対応 Exage/Storage

SINET4 Hiroshima University EXAGE L3VPN

SINET4 Kanazawa University EXAGE L3VPN

                       

proposed method shared NFS

Read (before migration) Read (after migration)Write (before migration) Write (after migration)

Throughput (MB/sec)

SC2013 2013/11/17~22 @Colorado Convention Center

We have been developing a widely distributed cluster storage system andevaluating the storage along with various applications. The main advantage ofour storage is its very fast random I/O performance, even though it provides aPOSIX compatible file system interface on the top of distributed cluster storage.

当初の予定

下條真司 Shinji Shimojo @Osaka Univ, NICT

面白く ないよね!

本番RTT=244ms 1Gbps

本番折り返し

国際回線を 国際回線上での DCダウン時の 今年は

82p/5min

Page 10: 4th RICC workshopのご案内

TOYAMA site

OSAKA site

TOKYO site before Migration

Copy to DR-sites

Copy to DR-sites

live migration of VM between distributed areas

real time and active-active features seem to be just a simple "shared storage". Live migration is also possible between DR sites

(it requires common subnet and fat pipe for memory copy, of course)

after Migration

Copy to DR-sites 広域分散 仮想化環境

Page 11: 4th RICC workshopのご案内

Distcloud

Page 12: 4th RICC workshopのご案内

DR Disaster Recovery

Page 13: 4th RICC workshopのご案内

1978

Page 14: 4th RICC workshopのご案内

Sun Information Systems

Page 15: 4th RICC workshopのご案内

mainframe hot site

Page 16: 4th RICC workshopのご案内

‘80-’90

Page 17: 4th RICC workshopのご案内

Realtime Processing

Page 18: 4th RICC workshopのご案内

POS point of sales

Page 19: 4th RICC workshopのご案内

’90-’00

Page 20: 4th RICC workshopのご案内

the Internet

Page 21: 4th RICC workshopのご案内
Page 22: 4th RICC workshopのご案内

2001.9.11 September 11 attacks

Page 23: 4th RICC workshopのご案内

2003.8.14 Northeast blackout of 2003

Page 24: 4th RICC workshopのご案内

in Japan

Page 25: 4th RICC workshopのご案内

2011.3.11 The aftermath of the 2011

Tohoku earthquake and tsunami

Page 26: 4th RICC workshopのご案内

BCP Business Continuity Plan

Page 27: 4th RICC workshopのご案内

群馬 Gunmma prefecture

Page 28: 4th RICC workshopのご案内

石狩 Ishikari city

Page 29: 4th RICC workshopのご案内

2つで 十分ですよ?

Page 30: 4th RICC workshopのご案内
Page 31: 4th RICC workshopのご案内

国立情報学研究所

Page 32: 4th RICC workshopのご案内

Kitami Institute of Technology

University of the Ryukyus

北見工大

琉球大学

SINET 最長

Page 33: 4th RICC workshopのご案内

XenServer 6.0.2

CloudStack 4.0.0

XenServer 6.0.2

CloudStack 4.0.0

Page 34: 4th RICC workshopのご案内

problems

Page 35: 4th RICC workshopのご案内

shared storage

Page 36: 4th RICC workshopのご案内

≒50ms

Page 37: 4th RICC workshopのご案内

RTT > 200ms

Page 38: 4th RICC workshopのご案内

分散ストレージ distributed storage

Page 39: 4th RICC workshopのご案内

要求性能 required quality

Page 40: 4th RICC workshopのご案内

64 256 1024 4096 16384 65536 262144 1.04858e+06 4.1943e+06 1.67772e+07 6.71089e+07 4 16

64 256

1024 4096

16384

0

20000

40000

60000

80000

100000

120000

Kbyt

es/s

ec

File size in 2^n KBytes

Record size in 2^n Kbytes

0

20000

40000

60000

80000

100000

120000

High Random R/W Performance

Page 41: 4th RICC workshopのご案内

POSIX準拠 interface protocl

NFS, CIFS, iSCSI

Page 42: 4th RICC workshopのご案内

���"���� �$����������� ��������������� �������!���� �

Con$idential �

�� ��� %*,&.'+�#�)(-���

Global VM migration is also available by sharing "storage space" by VM host machines. Real time availability makes it possible. Actual data copy follows.

(VM operator need virtually common Ethernet segment and fat pipe for memory copy)

TOYAMA site

OSAKA site

TOKYO site before Migration

Copy to DR-sites

Copy to DR-sites

live migration of VM between distributed areas

real time and active-active features seem to be just a simple "shared storage". Live migration is also possible between DR sites

(it requires common subnet and fat pipe for memory copy, of course)

after Migration

Copy to DR-sites

Page 43: 4th RICC workshopのご案内

Fileblock block block

block block block

block block block

Meta Data

consistent hash

backend (core servers)

Page 44: 4th RICC workshopのご案内

NFS CIFS iSCSI

Page 45: 4th RICC workshopのご案内

redundancy = 3

r = 2ACK

r = 1

r = 0

write

Page 46: 4th RICC workshopのご案内

redundancy = 3

ACK

r = 2 e = 0

r = 1 e = 0

r = 0 e = 1

r = -1 e = 2

external

Page 47: 4th RICC workshopのご案内

10Gbps

Cisco UCS

Hypervisor

VM

1/4U server x4

Page 48: 4th RICC workshopのご案内

大阪大学

金沢大学 広島大学

国立情報学研究所

Page 49: 4th RICC workshopのご案内
Page 50: 4th RICC workshopのご案内

iozone -aceI a: full automatic mode

c: Include close() in the timing calculations e: Include flush (fsync,fflush) in the timing calculations

I: Use DIRECT IO if possible for all file operations.

Page 51: 4th RICC workshopのご案内

write

64 256 1024 4096 16384 65536 262144 1.04858e+06 4.1943e+06 1.67772e+07 6.71089e+07 4 16

64 256

1024 4096

16384

0

20000

40000

60000

80000

100000

120000

Kbyt

es/s

ec

File size in 2^n KBytes

Record size in 2^n Kbytes

0

20000

40000

60000

80000

100000

120000

64 256 1024 4096 16384 65536 262144 1.04858e+06 4.1943e+06 1.67772e+07 6.71089e+07 4

16

64

256

1024

4096

16384

File size in 2^n KBytes

Record size in 2^n Kbytes

Page 52: 4th RICC workshopのご案内

0

20

40

60

80

100

120

10MB 100MB 1GB 10GB

Thro

ughp

ut (M

B/s

ec)

File size

write rewrite read reread

random read random write bkwd read

stride read fwrite freadlegend

record rewrite

0

20

40

60

80

100

120

10MB 100MB 1GB 10GB

Thro

ughp

ut (M

B/s

ec)

File size

0

20

40

60

80

100

120

10MB 100MB 1GB 10GB

Thro

ughp

ut (M

B/s

ec)

File size

0

20

40

60

80

100

120

10MB 100MB 1GB 10GB

Thro

ughp

ut (M

B/s

ec)

File size

0

20

40

60

80

100

120

10MB 100MB 1GB 10GB

Thro

ughp

ut (M

B/s

ec)

File size

0

20

40

60

80

100

120

10MB 100MB 1GB 10GB

Thro

ughp

ut (M

B/s

ec)

File size

0

20

40

60

80

100

120

10MB 100MB 1GB 10GB

Thro

ughp

ut (M

B/s

ec)

File size

0

20

40

60

80

100

120

10MB 100MB 1GB 10GB

Thro

ughp

ut (M

B/s

ec)

File size

0

20

40

60

80

100

120

10MB 100MB 1GB 10GB

Thro

ughp

ut (M

B/s

ec)

File size

0

20

40

60

80

100

120

10MB 100MB 1GB 10GB

Thro

ughp

ut (M

B/s

ec)

File size

0

20

40

60

80

100

120

10MB 100MB 1GB 10GB

Thro

ughp

ut (M

B/s

ec)

File size

従来方式 Exage/Storage

広域対応 Exage/Storage

Page 53: 4th RICC workshopのご案内

SINET4 Hiroshima University EXAGE L3VPN

SINET4 Kanazawa University EXAGE L3VPN

SINET4 NII EXAGE L3VPNSINET4 NII EXAGE L3VPN

Page 54: 4th RICC workshopのご案内
Page 55: 4th RICC workshopのご案内

                       

proposed method shared NFS

Read (before migration) Read (after migration)Write (before migration) Write (after migration)

Throughput (MB/sec)

Page 56: 4th RICC workshopのご案内

SC2013 2013/11/17~22 @Colorado Convention Center

Page 57: 4th RICC workshopのご案内

We have been developing a widely distributed cluster storage system andevaluating the storage along with various applications. The main advantage ofour storage is its very fast random I/O performance, even though it provides aPOSIX compatible file system interface on the top of distributed cluster storage.

㻯㼛㼚㼠㼍㼏㼠㼟㻦㻌㻰㼕㼟㼠㼏㼘㼛㼡㼐 㻼㼞㼛㼖㼑㼏㼠 㻱㻙㼙㼍㼕㼘䠖 㼐㼕㼟㼠㼏㼘㼛㼡㼐㻬㼞㼕㼏㼏㻚㼕㼠㼞㼏㻚㼚㼑㼠

• Long Distance: Sharing data across geographically dispersed locations• Multi-sites: Replicating data over at least three different locations• All Active: Simultaneous accessing from multiple locations

We have successfully performed a longdistance live migration experiment. Wehave migrated VMs using our storagewithout significant performancedegradation of read/write operations. Migrated to

remote siteMigrated to local site

Migrated to remote site

Migrated to local site

Fig. 1: Comparison of disk write performance during VM migration between with our platform and with NFS. (Distance between two sites is about 450km and RTT is about 18ms.)

Page 58: 4th RICC workshopのご案内

当初の予定

Page 59: 4th RICC workshopのご案内

下條真司 Shinji Shimojo @Osaka Univ/JGN-X Leader

Page 60: 4th RICC workshopのご案内

面白く ないよね!

Page 61: 4th RICC workshopのご案内

本番

Page 62: 4th RICC workshopのご案内

RTT=244ms 1Gbps

本番折り返し

Page 63: 4th RICC workshopのご案内

国際回線を 使用した

マイグレーション

Page 64: 4th RICC workshopのご案内

国際回線上での 広域分散ストレージのアクセス試験

Page 65: 4th RICC workshopのご案内

DCダウン時の DR実現検証

Page 66: 4th RICC workshopのご案内

今年は 拠点を 米国に

Page 67: 4th RICC workshopのご案内

Future Works

Page 68: 4th RICC workshopのご案内

Big Data Analysis

Page 69: 4th RICC workshopのご案内

モバイルデバイス からの行動データ behavior data from mobile devices

Page 70: 4th RICC workshopのご案内

電源非供給地域で 収集されるデータ data from non-electrification area

Page 71: 4th RICC workshopのご案内

personal data aggregation service

high latency power

consumption

mobile devices

sensor devices

Page 72: 4th RICC workshopのご案内

mobile devices

sensor devices

low latency

regional exchange

regional exchange

personal data aggregation service

wide-area distributed platform

regional data center

regional data center

Page 73: 4th RICC workshopのご案内

経路最適化

Page 74: 4th RICC workshopのご案内

【今後の展開】仮想計算機の流動性向上に向けて

18

VM VM VMmigration

Page 75: 4th RICC workshopのご案内

拠点間マイグレーションにおける経路最適化の実現

21

Layer� e�� �¶� �[ŕ!�œ��Ŗ�

L3�

¦À"Z�œæŅĚĤŏŔĪŋŒĕĵōġą¦ÀQ1ĕ&�ą�đxò�

ĨĚĵ)�ĂƦÀ"Zô+¯ VM)�ĂĆ|ÁĄ"Zô6Þ ¦Àl�ô\¶ąĄĒ�

+�

L2S��œæY{ĆL2f³ĕ�� œæVPLS, IEEE802.1ad PB(Q-in-Q),

IEEE802.1ah(Mac-in-Mac)

FG^íğłŏŔĪŋŒħĬĵą�ēĒ h�pąKûĒĬĥŔŌĻōiję (Ç%ħĬĵíľŐŔĶĢňĬĵĶņĚŒ�Ŗ

L2 over L3� œæĝŒĶŃĬĵÖĂIPĵŒķōŒĤ œæVXLAN, OTV, NVGRE�

tDL3f³ĂH�+¯ ĵŒķōŒĤğŔĹŔŀIJĶ IPŅŎıĢňĬĵô\à�

SDN�œĴŔİĆĽĞőŔĴęŒĤĕįĽĵěĜĘĂ"Z œæOpenFlow

ĿŐĤŌŅľŎ"Z �3Ç%íÎ�ħĬĵå

ID/Locator�Ý�œĸŔĶĆIDăŎŔijęŒĤ�ĆLocator ĕ�Ý œ LISP�

h�pąKûĒĬĥŔŌĻōiję ĸŔĶ)�Ć"Z �3Ç% ŅIJļŒĤ_<¡�í*u^ąWß�

IPŇĻōijęf³� œ IPMĂĆ�&ÉÏ^ œæMAT, NEMO, MIP(Kagemusha)�

h�pąKûĒĬĥŔŌĻōiję IPMĂĆ��¼°(�¯ĕAÒĂH�ûĒe�čD8Ŗ�

L4� mSCTP� œ SCTPĆŅŎıĺĬ�¯ĕ!� L2 / L3ąć×�ùĄð ĵŌŒĬńŔĵôSCTPąÙGøēĒ�

L7� DNS + Reverse NAT�œ Dynamic DNSąďĒ&�Ą�đxò œ VMąćĿŌĚŁŔĵĘĶŏĬĕ��ùāíh��ĆĦŔĵěĜĚĂReverse NATĕÑ��

L2 / L3ąć×�ùĄð IPĘĶŏĬć@v ĮIJĪŋŒć�røēĒ�

Page 76: 4th RICC workshopのご案内

2011.3.11 The aftermath of the 2011

Tohoku earthquake and tsunami

Page 77: 4th RICC workshopのご案内

Japan

Taiwan

Indonesia

New Zealand

Page 78: 4th RICC workshopのご案内

4th RICC workshop @Okinawa 2014/3/27(Thu)~28(Fri)

Page 79: 4th RICC workshopのご案内

http://ricc.itrc.net

Page 80: 4th RICC workshopのご案内

Cybermedia Center Osaka UniversityCybermedia Center Osaka UniversityCybermedia Center Osaka University

そういえば任期が 2014/3/31まで

なので…

Page 81: 4th RICC workshopのご案内

転職先 探しています

Page 82: 4th RICC workshopのご案内
Page 83: 4th RICC workshopのご案内
Page 84: 4th RICC workshopのご案内

おあとが よろしい ようで

Page 85: 4th RICC workshopのご案内

go to next stage