1. IBM FlashSystem…IBM FlashSystem…
Enterprise Storage in microseconds, not millisecondsEnterprise Storage in microseconds, not milliseconds
Accélérez votre Business…Accélérez votre Business…
Maintenant et facilement!Maintenant et facilement!
Proposition de Valeur
Solutions et implémentation
Bénéfices clients
Contact:
Pierre-Jean BOCHARD pjard@fr.ibm.com – 06 88 21 98 76
http://www-03.ibm.com/systems/storage/flash/index.html
2. Le portefeuille stockage IBM
IBM Smart
Analytics
Solutions
(ISAS)
Solutions intégrées
Haut de gamme
XIV SONASDS8000
Midrange
et entrée
de gamme
Protection
et conservation
des donnés
Systèmes de stockage optimisés
Tape Library
TS3310
Tape Virtualization
TS7720/7740
Tape Automation
TS3500
Tape drives
Enterprise & LTO
ProtecTIER
TS7620/50
Tivoli Storage Productivity Center
Tivoli Storage FlashCopy
Manager
Tivoli Storage Manager
Tivoli Key Lifecycle Manager
Logiciels
de gestion
du stockage
Innovation
intégrée
Storage
Virtualization
SW and SVC
Real-time
Compression
Déduplication
Easy Tier
IBM Active
Cloud EngineTM
Linear Tape File
System (LTFS)
Virtual Storage Center
Storwize
V3700
DS3500
DCS3700N series
Storwize V7000
& Storwize V7000 Unified
2
SmartCloud Storage Access
Flash Storage
Systems
FlashSystems
3. Durant les 10 dernières années:
• Vitesse CPU: croissance 8-10x
• Vitesse DRAM: croissance 7-9x
• Vitesse BUS: croissance 20x
• Vitesse RESEAU: croissance 100x
• Vitesse DISQUES: croissance 1.2x seulement
Le disque, goulot d’étranglement dans l’IT
4. 1. Multiplier par 10 l’efficacité de vos applications… Oracle, SAP, BI, ….
2. Multiplier par 10 l’efficacité de vos serveurs…
3. Augmenter l’efficacité de vos opérations (backup, infra VDI,…)
4. Réduire potentiellement les coûts de vos licences ISV?
5. Réduire vos coûts d’hébergement, de bilan énergétique par 10 et +
SANS RE-ARCHITECTURER VOS APPLICATIONS,SANS RE-ARCHITECTURER VOS APPLICATIONS,
SANS CHANGER VOS SERVEURSSANS CHANGER VOS SERVEURS
Avec un ROI de quelques moisAvec un ROI de quelques mois
Que diriez vous si vous pouviez…
5. ‘FLASH will change the structure of our future systems…
…Texas Memory Systems, leader in FLASH OPTIMIZED SYSTEMS…
We will be able to offer our clients flash optimized systems
by integrating this technology in our overall portfolio’
Ambuj GOYAL
GM, System Storage & Networking
Systems & Technology Group
• Les plus grosses installations de Flash en production dans le monde
• En opération dans 10 places financières majeures dans le monde
• Utilisé par 7 des 11 plus gros opérateurs telcos dans le monde
• Installés et en production dans 35 pays
• Envoyer un texto
• Parier en ligne
• Réserver un vol, une croisière
• Utiliser un GAB
• Trader en ligne
• Acheter en ligne
• Utiliser du wifi pre-payé
• Jouer en ligne
TMS Flash Systems are everywhere
…where Performance is Key for Business!
6. Proposition de valeur IBM FlashSystem
85% Reduction
In batch processing
times
85% Reduction
In batch processing
times
90% Reduction
In OLTP times
90% Reduction
In OLTP times 150-200 µs Latency150-200 µs Latency
80% Reduction
Energy Usage
80% Reduction
Energy Usage
75% Footprint
Reduction
Store one petabyte
in a single floor tile.
Add compression
and add up to 100%
more
Enterprise
Reliability
High Availability,
2D Flash RAIDTM
and Variable
Stripe RAIDTM
7. Faire un I/O: Baie de disques actuelle
Time
I/O Serviced by Disk
1. Issue I/O request (~ 100 μs)
2. Wait for I/O to be serviced (~ 5 000 μs)
3. Process I/O (~ 100 μs)
Time to process 1 I/O request = 200 μs + 5 000 μs = 5 200 μs
Server: CPU Utilization = Wait time / Processing time = 200 / 5 200 = ~4%
Processing
~100µs ~100µs
Waiting
~5000 µs
CPU State
of server
1 I/O Request
8. Faire un I/O: Baie de disques avec des SSD
Time
I/O Serviced by SSD
1. Issue I/O request (~ 100 μs)
2. Wait for I/O to be serviced (~ 1 000 μs)
3. Process I/O (~ 100 μs)
Time to process 1 I/O request = 200 μs + 1 000 μs = 1 200 μs
Server: CPU Utilization = Wait time / Processing time = 200 / 2 200 = ~17%
Processing ~100µs ~100µs
Waiting ~1000 µs
CPU State
Of Server
1 I/O Request
Flash dans un format
disque: Solid State Drive
(SSD)
Problémes:
1/ Limitation des IO/sec
due à la saturation des
controleurs et bus
internes
2/ Financier:
> 6 SSD Flash TMS
9. Faire un I/O avec une baie IBM FLASH:
Time
I/O Serviced by Flash Memory
1. Issue I/O request: ~ 100 μs
2. Wait for I/O to be serviced: ~ 200 μs (vs 5000 μs HDD)
3. Process I/O: ~ 100 μs
Time to process 1 I/O request = 200 μs + 200 μs = 400 μs
Utilisation CPU du serveur : Wait time / Processing time
200 / 400 = ~50% (versus 4% avec un I/O disque)
Processing
~100µs ~100µs
Waiting ~200 µs
CPU State
Of Server
1 I/O FLASH
Un facteur 12
L’application a ses
performances multipliées
par 12 avec la Flash !
CPU plus utilisée
moins de coeurs
budget licences ISV
réduit (Oracle, SQL)
~5000 µs
~100µs
1 I/O Disque
10. Positionnement des systèmes de disques IBM
Serveurs IBM Mainframe Serveurs Open
Performances(Resp.Time,IOps)
IBM Flash TMS
Reliability,Availability,Serviceability
Block - SAN FC
System Z System P / I / PureSystems
Unified:SAN+NAS
UltraHighHighMedium
microseconde
DS8870 SoNAS
SVC + DS8870 /V7000/XIV
Storwize
Family
Storwize
V7000
Unified
N series
Entry
XIV
(One Tier/Grid)
milliseconde
NAS
SOD
12. Offre de baies FLASH IBM Pour toute taille de client
De 1 TeraOctet dans 1 U à 1 PetaOctet dans 1 seul Rack!
De 450 000 I/O à 20 millions d’I/O par seconde!
De 25 à 110 micro secondes de temps de réponse! (*)
(*) Temps de réponse classique d’une baie de
disques: 5 millisecondes
FlashSystem 820 (eMLC) FlashSystem 720 (SLC)
FlashSystem 710 (SLC)FlashSystem 810 (eMLC)
13. Baie Flash 720 / 820 – A coeur ouvert
1 U, 500Watts, 500 000 IOPS, latence: 25 (read) à 110 (write) microsecondes
Redundant
Power Supplies
Redundant Fans
12 Flash Modules (10+1+1)
1U Chassis
N+1 batteries
Redundant
Management
Control
Processors
Redundant
RAID controllers
Two Dual-Ported 8 Gb/s FC or
40 Gb/s QDR IB Interfaces
14. RAID 5 across Flash Modules (10 data + 1 parity + 1 hot spare)
External
Interfaces
(FC, IB)
RAID
Controllers
RAID 5 within
Flash Modules
(9 data + 1 parity)
TMS
2D Flash RAID™
Protection des data - 2D Flash RAID™
15. 3 modes d’implémentation d’une baie IBM Flash / TMS…
… sans modifier l’infrastructure applicative existante
17. Exemple d’infra en High Avalaibility avec
SVC Stretched Cluster + IBM Flash / TMS
Room 1
Blade
Center 1
Blade Blade
IBM Flash
5TB
SANSAN
Storwize
V7000
SSD
SAN SAN SAN SAN
SVC
Stretched cluster
Node 1
Blade Blade
SANSAN
SVC
Stretched cluster
Node 2
Storwize
V7000
HDD
IBM Flash
5TB
Storwize
V7000
SSD
Storwize
V7000
HDD
Room 1
Blade
Center 2
19. Au niveau du SAN
SVC / V7000 Primary Copy of vDisk
Au niveau de l’OS (Volume Manager)
AIX LVM (native least queue read)
Veritas (Preferred Read Plex)
Solaris SVM
Oracle ASM
Au niveau applicatif
Standby/Reporting Instance
SQL AlwaysOn
Implementation Flash en parallèle du stockage existant
23. • COCC, financial
technology provider
• Custom Oracle ATM
processing application
• Mirrored TMS hybrid
Flash array
• CIO Chad Burney won
InfoWorld CTO 25 award
for performance and cost
improvements
For Internal Customer Use Only - Texas Memory Systems Confidential
24. Enterprise Array, No FlashEnterprise Array, No Flash IBM Flash / TMSIBM Flash / TMS
2 million queries
12.25 minutes to complete
16K Total IOPS
4K per RAC Node
[oracle]$ time ./spawn_50.sh
real 12m15.434s
user 0m5.464s
sys 0m4.031s
2 million queries
12.25 minutes to complete
16K Total IOPS
4K per RAC Node
[oracle]$ time ./spawn_50.sh
real 12m15.434s
user 0m5.464s
sys 0m4.031s
2 million queries
1.3 minutes to complete
160K Total IOPS
40K per RAC Node
[oracle]$ time ./spawn_50.sh
real 1m19.838s
user 0m4.439s
sys 0m3.215s
2 million queries
1.3 minutes to complete
160K Total IOPS
40K per RAC Node
[oracle]$ time ./spawn_50.sh
real 1m19.838s
user 0m4.439s
sys 0m3.215s
A factor of over 9x improvement!
Performance Scenario: Oracle RAC, 4 Nodes
25. Problem
•UFC provides digital content for over 130
platforms, transcoding created storage
bottlenecks
Solution
•5TB of RamSan-710, equivalent to the
performance of 2,200 15K RPM hard disk
drives
Benefit
•70% reduction in transcoding time
•40% reduction in delivery time to 130
platforms
Problem
•UFC provides digital content for over 130
platforms, transcoding created storage
bottlenecks
Solution
•5TB of RamSan-710, equivalent to the
performance of 2,200 15K RPM hard disk
drives
Benefit
•70% reduction in transcoding time
•40% reduction in delivery time to 130
platforms
“The RamSan has just completely
elminated bottlenecks. Everything is
flowing smoothly” Danny Gold – UFC
UFC Case Study
Video production and distribution
26. The Research Computing and Cyberinfrastructure (RCC) group at Penn State University provides
systems and services that are used extensively in computational and data-driven research and
teaching by more than 5,000 faculty members and students.
The situation Research initiatives slowed by HDD
To deliver on that mission, RCC operates computational clusters composed of 1,000 servers with
10,000 CPU cores.
RCC’s storage infrastructure consists of a cluster of 14 servers running Red Hat Enterprise Linux 5.
The servers are connected through 8 Gbps Fibre Channel HBAs to 900 TB of disk storage on a
Storage Area Network (SAN) with that data served out over Ethernet to the computational clusters.
The RCC storage cluster is built on General Parallel File System (GPFS) of IBM®, a highly scalable
clustered file system that leverages file metadata to achieve policy-based active management of
files for backup, migration, archiving and index tagging of files.
It also places an extreme random I/O workload on the storage infrastructure by having to process
metadata for those files on an ongoing basis.
The GPFS metadata files are small—only 512 bytes—so capacity is not the issue. Rather, the IOPS
requirements to process millions of files on an
an ongoing basis were overwhelming Penn State RCC’s SAN infrastructure.
HPC infra at Penn University
27. The challenge Speed up nightly backups
RCC had allocated two hundred 15K RPM hard disk drives to handle
the metadata processing that managed the nightly backup of user data.
The group had to overprovision capacity of the drives in an attempt to
generate acceptable IOPS to complete the backup operation in a brief
overnight window so as not to impact production processes.
Despite allocating a massive number of the disks to the operation, the
nightly incremental backup were still taking up to 6 hours and
degrading other system operations.
28. The solution High-performance, reliable RamSan-810 flash storage
The team at RCC determined that a solid state storage solution was needed to
effectively handle the IOPS load and replace the HDDs, which had been over
provisioned in an attempt to boost IOPS.
“We didn’t need the capacity but we did need the performance,” said Vijay
Agarwala, Sr. Director for Penn State’s RCC.
““There were operations, such as nightly backups which have to check the
status of all the files in the system, that were being bottlenecked by the
performance of those disks.
So it made sense to go with a Flash media array rather than continuing to
throw traditional disk spindles at the problem,” said Jason Holmes, lead
systems administrator at RCC.
29. “GPFS allows us to move the
metadata from the disk to the
RamSan online and once we did that
the backups were reduced down to
about an hour. So that was a huge
difference,” said Holmes. “TMS for
this application was the best solution
largely because of its maturity and
performance. It seemed very stable
and it just worked out of the box.
The applications usage at RCC consists of an 85/15 read-write ratio,
so eMLC offered a better value than the increased write performance
—but higher cost—of a Single Level Cell system, such as the
RamSan-710.
FlashSystem implementation
30. The result Instant six times performance boost
“With some of the other solutions we tested, we poked and pried at them for
weeks to get the performance where the vendors claimed it should be,” said
Michael Fenn, systems administrator at Penn State.
“With the RamSan we literally just turned it on and that’s all
the performance tuning we did. It just worked out of the box.”
“GPFS allows us to move the metadata from the disk to the RamSan
online and once we did that the backups were reduced down to about
an hour. So that was a huge difference,”
“TMS for this application was the best solution largely because of its
maturity and performance. It seemed very stable and it just worked out
of the box.
Easy to implement, immediate benefits
31. Another thing we’ve seen is increased responsiveness in the file system
with users, and commands like listing large directories run five times as
fast as they did previously. The IOPS capacity of the RamSan has
significantly speeded those up.”
Additional benefits
Swapping 200 power-hungry 15K hard disks for two RamSan-810s provided yet
another benefit for RCC, reducing its power consumption for metadata storage in the
GPFS file systems by 90 percent, from about 5000 watts down to just 500 watts.
The power savings are just part of the cascading cost savings that RamSan Flash
storage creates. RamSan Flash storage systems not only require less power to
operate, but also produce less heat and consumes less rack space, lowering data
center cooling and floor space costs.
32. Problem
•Nightly back up took 6 hours;
degraded system performance
Solution
•Implemented 4TB of
RamSan-810
Benefit
•Backup window improved 6x to one
hour
Problem
•Nightly back up took 6 hours;
degraded system performance
Solution
•Implemented 4TB of
RamSan-810
Benefit
•Backup window improved 6x to one
hour
“With the RamSan we literally just turned
it on and that’s all the performance tuning
we did. It just worked out of the box”.
Jason Holmes – Penn State University
Penn State Case
Study
HPC / Computational Applications
33. Une solution IBM mature, opérationnelle chez plus de 300 clients
34. Ou souhaitez vous accélérer votre business?
Accélérer votre business rapidement, avec un ROI court
(90% bases SAP < 2TO) Solutions Business Intelligence IBM
35. Infrastructure: Accélérer les applications sans tout changer
Améliorer les perfs sur les applis business gourmandes en IO
• Satisfaire les utilisateurs en un temps rapide et avec ROI court
• Alignement plus rapide de l’IT sur le Business
• Eviter de changer toute une infrastructure Risque, côuts, durée
• Optimisation potentielle de côuts de license SW
• Réduction potentielle du nombre de cœurs sur les serveurs
Intégrer aisément dans l’existant
• Peu de changement opérationnel
• Amélioration des opérations (chaines batch, backup)
• Réduction potentielle des heures
Mieux utiliser le stockage en place
• Optimisation durée de vie existant
• Résolution pb de budget
Résoudre des problématiques de salle machine
• Ajout de Flash / Disques espace, power, cooling maitrisés et réduits
• Budget maitrisé si infra en location chez un hébergeur
Big Data, Analytics, ORACLE, SAP, HPC, VDI, Backup
36. Application: Architectes/Propriétaires,
Line of Business Managers
DSI
• Réduit les plaintes des users
• Supporte plus d’utilisateurs
• Réduit drastiquement les temps de
réponse
• rapide à MEO, sans changement applicatif
• Les meilleurs indicateurs: ROI rapide, TCO
abaissé, économie potentielle de licence SW
• Faible consommation électrique, dans 1 U
Administrateurs de base de données
Responsible d’Infrastructure
(Serveur ou Storage )
• Suprime les I/O waits
• Réduit les temps de ‘query’
• Accélère les applications
• Evite d’entrer dans une architecture
‘vérouillée’ par le fournisseur (vs. Exadata)
• Integration sans peine dans l’infra existante
• Faible consommation électrique, dans 1 U
Proposition de valeur d’IBM Flash
37. Line of Business, propriétaires de processus métier:
• Quelles sont vos applications les plus critiques dans votre secteur d’activité?
• Une amélioration d’un facteur 2 minimum de ces applications (temsp de réponse, nombre
d’utilisateurs) aurait quel impact sur vos résulats?
• Avez vous une Stratégie Flash?
CIO, CTO, DBA:
• Vos utilisateurs se plaignent ils d’un manque de performances?
• La performance limite t elle vos objectifs business et/ou l’atteinte de vos niveaux de service
• Connaissez vous le taux d’utilisation CPU de vos serveurs?
• Comment mesurez vous la performance de vos applications métiers? De vos infrastructures?
• Quelle est votre stratégie pour améliorer les performances?
• Cela va t il nécessiter de remettre à plat l’architecture logicielle? Et l’architecture Serveurs /
Stockage?
• Avez vous une problématique de coûts de licences (Oracle, Citrix, Vmware)?
• Avez vous une probématique d’espace en salle machine, de consommation éléctrique et de
refroidissement?
• Quelle performance aurez vous besoin dans un an?
Une solution IBM Flash apporte t elle de la valeur
dans votre environnement business et IT?
38. go to http://www.redbooks.ibm.com and search for IBM FlashSystem
IBM FlashSystem 710 and IBM FlashSystem 810:
http://www.redbooks.ibm.com/technotes/tips1002.pdf
IBM FlashSystem 720 and IBM FlashSystem 820:
http://www.redbooks.ibm.com/technotes/tips1003.pdf
IBM FlashSystem in OLTP Database Environments:
http://www.redbooks.ibm.com/technotes/tips0973.pdf
IBM FlashSystem in OLAP Database Environments:
http://www.redbooks.ibm.com/technotes/tips0974.pdf
Configurations dans outil IBM e-config
Support avant-vente: Bruno Sourbe
Email: bruno_sourbe@fr.ibm.com
Mobile : 06.73.84.57.44
Benchmark: IBM Montpellier
Plus d’informations
39. Accélérer votre Business …
C’est maintenant!
Avec les solutions
IBM FlashSystem
http://www-03.ibm.com/systems/storage/flash/index.html
Merci pour votre attention
Editor's Notes
On this slide we have the rate of performance improvement by element in the data path over the last 10 years. As you can see spinning disk is clearly the bottleneck in today’s hardware stack – and, to make matters worse, the gap is widening over time. Disk will trail other elements to a greater degree in the future. It would logically follow, if disk speed is the slowest element in the chain, it determines how quickly an application can access information. And the application determines how quickly an end used can make a decision, respond to a customer or get product out the door. In effect, disk response time determines the pace at which a business operates. Admittedly, that’s a little over simplified, but you can see where I’m going. Two key questions that come to mind are: is their pace aligned with the pace of their industry? If that pace can be improved, will it create competitive advantage? Now for you with a more financial bent, let’s examine this strictly from a financial perspective. The most efficient use of a corporation’s finite capital allocation is to address the piece in the technology stack that is fettering the rest of the infrastructure. Injecting Flash into an existing disk farm will have the greatest impact per dollar spent – because in addition to directly improving storage asset utilization, there is a ripple effect, enabling the other assets in the data center ecosystem to be more efficiently utilized. Because the other components can now do more work. It essential unlocks the value of all the money previously spent on IT. As you will see illustrated in the next two slides, the proverbial CPU can, in a sense, wait faster. . .
{TRANSCRIPT} Let’s quickly walk through an IO flow with spinning disk. Please note this is not a vendor specific thing. This concept applies to all spinning disk. Important note: These timings are approximate, and for a specific architecture and workload. They are meant to illustrate an example only. Note: The ~100us wait times are non-storage related and therefore independent. This example shows the case when a read request is not satisfied from cache in a disk array. In cached disk array, some portion of read requests will be satified out of cache. Step through the build click by click.
{TRANSCRIPT} Now let’s look at the same flow when our RamSan has been substituted for spinning disk. Walk through build. Once again, this example shows the case when a read request is not satisfied from cache in a disk array. In cached disk array, some portion of read requests will be satified out of cache. Improving application performance by reducing cpu wait time by a factor of up to 12x. Note, there was no tuning, tiering or software added. Simply plug it in an get the benefit. That my friends is “why Flash”. So let’s dive into the technology itself. What is flash memory?
Que devrions-nous faire, comment et que peut-il se passer ? Business intelligence Connecter vos collaborateurs aux informations au moment et à l'endroit où ils en ont besoin et de la façon dont ils le souhaitent à l'aide de fonctionnalités de reporting, de tableaux de bord, et d'outils d'analyses temps réel. Analyse prédictive Intégrer l'analyse prédictive dans le pilotage de la performance vous permet d'accroître la visibilité de votre activité en capturant toutes les informations dont vous avez besoin sur les attitudes et opinions de votre clients, prévoir les résultats et interactions avant qu'ils ne se produisent. Financial Performance Management Disposer de plans et prévisions continues basés sur les moteurs clés de l‘activité vous aident à mieux allouer vos moyens humains et financiers, de simuler plusieurs hypothèses, de vérifier l'impacts de vos décisions sur votre plan de marche. Pilotage du risque Approfondir la connaissance dans tous les aspects de la gestion du risque (GRC) avec une solution intégrée et faciliter l'adoption des meilleurs pratiques en terme de conformité. Applications analytiques Rentabilisez vos investissements informatiques existants à l'aide de rapports, tableaux de bord et analyses prêts à l'emploi et répondant aux meilleures pratiques du marché.
Fiabilité: Haute dispo des données assurée par la redondance des controleurs Protection raid sur la mémoire flash: 1ere dimension: RAID dans une carte chips flash + 2eme dimension: RAID parmi les 20 cartes dont 1 de spare automatique Variable Strip RAID: