3. Cloud Computing?
National Institute of Standards and Technology definition is:
• Cloud computing is a model for enabling convenient, on-demand network
access to a shared pool of configurable computing resources (e.g., networks,
servers, storage, applications, and services) that can be rapidly provisioned
and released with minimal management effort or service provider interaction.
Key Cloud Layers Deployment
Characteristics • Applications Models
• Rapid provisioning • Platform
• Private Cloud
• High elasticity • Infrastructure
• Public Cloud
• Access anywhere, any time • Hybrid Cloud
• Resource pooling
• Pay as you go
6. Innovation to solve the typical Bottlenecks
Problems?
Addressing SAN bottlenecks Massive Parallel Grid
Dumb SAN of DB Content with ultra- fast intelligent Storage
Database Exadata Storage
Database Traditional
Servers Servers
Servers SAN
6
8. Engineered Systems & Appliances
Datacenter Transformation Through Innovation
Purpose Built General Purpose
Exadata Exalogic ZFS Storage Appliance
SPARC
SuperCluster
Database Appliance Exalytics
9. SPARC Super Cluster
Hardware and Software
Engineered to Work Together
• MAA for General purpose
• Flexible for DataBase, MiddleWare &
Applications
• Optimized and integrated for better
performance, reliability, security,
management
• Reduced change management risk
• Lower cost of ownership
10. SPARC Supercluster
SPARC T4 Servers
Redefining General Purpose • 12 World Records over IBM
and HP across every tier
• Built-in Crypto
Exadata Storage Server
• Intelligent Storage Grid InfiniBand NetWork
• Hybrid Columnar Compression
• 40 GB/s for Storage & Nodos
• Smart Cache Flash connectivity
• 1.0M IOPS, 32 GB/s throughput • 5-8x the speed of current
networks
Exalogic Elastic Cloud ZFS Storage Appliance
• Active GridLink for RAC • Hybrid Storage Pools
• Extreme Java Performance 10x • DTrace Analytics
• JDBC over SRP • Built-in Data Services
Solaris 11 Integrated Management
• OVM Live Migration
• Compatibility Solaris 8,9 &10 • Enterprise Manager (opt.)
• Cloud provisioning in seconds
• Near zero virtualization • EM Ops Center (opt.)
overhead • ILOM
11. SPARC SuperCluster Architecture
• Best infrastructure solution for enterprise applications
• 1,200 CPU threads
Exadata Storage Servers • 4 TB DRAM
• 97 to 198 TB Hard Disk
SPARC T4-4 Compute Nodes
ZFS Storage Appliance
• 8.66 TB Flash
InfiniBand Switches
• 1.2M IOPS
• 42 GB/sec Storage Bandwidth
• 896 Gb/sec InfiniBand
Interconnect
14. SPARC T4 Java & Database Performance
SPECjEnterprise2010
2.4x faster than Power7 & DB2 and WebSphere
7x better price performance for Java
IBM P780 T4-4 Servers
$1,297,956 $467,856
See benchmark attribution slide.
15. SPARC T4-4 World Record TPC-H 1TB
Beats IBM’s Claims of 4:1 Core Performance Advantage
$800K cheaper and 22% faster than Power7 & Sybase
$125K cheaper and 3.6x faster than HP Superdome & Oracle 11g
TPC-H @1000GB
201,487
164,747
140,181
QphH
As of Dec 16, 2011 See benchmark substantiation slides
17. SPARC SuperCluster Architecture Diagram
Customer 10GBE Network User
T4-4 Node 1 T4-4 Node 2 T4-4 Node 3 T4-4 Node 4
Solaris 11 Solaris 10 Solaris 10 Solaris 11
GP Domain GP Domain GP Domain GP Domain
Solaris 11 Solaris 11 Solaris 11 Solaris 11
Exadata Dom Exadata Dom Exadata Dom Exadata Dom
InfiniBand Network
ZFSS
EXADATA EXADATA EXADATA STORAGE
ZFS STORAGE
APPLIANCE
STORAGE STORAGE STORAGE AAPPLIANCE
18. Exadata & SuperCluster Storage Servers Innovations
Up to 10–50X Performance Improvements
Intelligent storage Hybrid Columnar Compression
– Smart Scan query offload – 10x compression for warehouses
– 15x compression for archives
– Scale-out storage
Data remains
+ + + Uncompressed
compressed
for scans
Smart Flash Cache and in Flash
– Accelerates random I/O up to 30x
primary backup
– Doubles data scan rate
Benefits test
Multiply standby dev
Compressed
19. Exalogic Elastic Cloud
Exabus – High Speed Network Virtualization
Direct Memory Java I/O for Exalogic & SPARC SuperCluster
Standard Hardware I/O Exabus
Application Application Application Application
Application Buffer Application Buffer
Copy TCP IP Transport
Kernel
20% Buffer Copies Zero Buffer Copy
40% Transport Processing Direct Memory Access
40 % Kernel Context Switches Kernel Bypass
4X Throughput, 6X Lower Latency
19
21. Oracle Solaris 11
Mission Critical Meets Cloud
Designed-in Scalable Advanced Simplified
Virtualization Data Management Protection Administration
• Fully virtualized: • 10x storage • Immutable root • 3x faster
OS, network and savings for file system provisioning
storage virtualization vs VMware
• Virtual network • 2x storage • 4.3x faster • 4x faster
services built in compression OpenSSL upgrades
vs AIX vs Red Hat
• 15x lower • 2x SPC-1 IOPs
virtualization vs NetApp at ½ • 2.5x faster
overhead the cost • 3x faster ZFS reboots
vs Vmware encryption on vs Red Hat
SPARC vs
• 4x lower latency Westmere x86 • 46% fewer
vs. KVM hardware outages
using FMA
2,700 Projects 400 Unique Innovations 56 Million Tests
25. Consolidate on SPARC SuperCluster
Simplify , Accelerate, and Reduce Operating Expenses
• Consolidate up to 100 legacy
midrange servers to 1 SPARC
SuperCluster
ERP
• Fewer, smarter processors
CRM dramatically reduce:
• Software licensing costs
• Support costs
HR • Power and cooling
• Floor space consumption
BI • Run a wide range of new and
existing databases and multi-tier
eCom
applications on a single system
26. SPARC SuperCluster Virtualization
T4-4 Node 1 T4-4 Node 2 T4-4 Node 3 T4-4 Node 4 • Built-in, low overhead
Solaris 11 Solaris 10 Solaris 10 Solaris 11 virtualization
GP Domain GP Domain GP Domain
• Optimized Exadata
GP Domain
domains
Solaris 11 Solaris 11 Solaris 11 Solaris 11
• General Purpose
Domains support full
Exadata Dom Exadata Dom Exadata Dom Exadata Dom
range of workloads
InfiniBand Network
• Runs Solaris 10 and 11
Exa
data
Exa
data
Exa
data
Exa
data
Exa
data
Exa
data ZFSSA • Simultaneously run
multiple databases,
applications and tiers
34. Fundamental Value of Engineered Systems
Easy, Predictable, Just Works
Cut Deployment
Times by up to
75%
High Availability
Built in
99.999% Uptime
100’s of Components 1 Machine
1000’s of Hours 1 Day
36. Oracle Premier Support For Full Stack
Complete Coverage
Service and Tools and Product
Support Resources Innovation
Quickly diagnose Get the most of your Keep pace with change
and resolve issues Oracle products with and capitalize on
proactive services new opportunities
38. SPARC SuperCluster
Unmatched Business Value
Extreme • Run existing applications up to 4x – 8x faster
• Up to 10x to 50x database acceleration with Exadata Storage Servers
Performance
• 4x – 8x better price/performance than competing servers
Complete • TCO savings of up to 5x over 5 years
Lower TCO • Cut deployment time and administration costs up to 75%
• Dramatically reduce software licensing and administrative overhead
• Consolidate up to 100 legacy servers to 1 SuperCluster with no software
Simplified migration costs
• Built-In virtualization, security and systems management
Consolidation • Oracle Solaris offers guaranteed investment protection
38
39.
40. Inspiration from Everyday Life
• Modular design
simplifies
production
• Allows
differentiation
through colours,
character themes
and packaging
41. SPARC SuperCluster
Business Value
Extreme • Optimized performance for Full Stack Applications
Performance • Based on world record breaking SPARC T4-4 servers
• Exadata Storage & Exalagic Cloud performance
Complete
• Reduced storage requirements
Lower TCO • Accelerates deployments
• Leverages existing investments
Datacenter • Runs a mix of Solaris 10 and 11
• Runs multiple databases
Consolidation • Runs multiple application tiers
41
Notes de l'éditeur
Multiple definitions of cloud – pulled from two places:Wikipedia and National Institute of Standards and Technology (NIST) Definition of Cloud Computing: Internet based computing, shared resources, delivered on demand like a public utility. Those shared resources can be networks, servers, storage, applications and services. Over the internent requires less model. Essential Characteristics:On-demand self-service. A consumer can unilaterally provision computing capabilities, such as server time and network storage, as needed automatically without requiring human interaction with each service’s provider. Broad network access. Capabilities are available over the network and accessed through standard mechanisms that promote use by heterogeneous thin or thick client platforms (e.g., mobile phones, laptops, and PDAs). Resource pooling. The provider’s computing resources are pooled to serve multiple consumers using a multi-tenant model, with different physical and virtual resources dynamically assigned and reassigned according to consumer demand. There is a sense of location independence in that the customer generally has no control or knowledge over the exact location of the provided resources but may be able to specify location at a higher level of abstraction (e.g., country, state, or datacenter). Examples of resources include storage, processing, memory, network bandwidth, and virtual machines. Rapid elasticity (dynamic capcity) . Capabilities can be rapidly and elastically provisioned, in some cases automatically, to quickly scale out and rapidly released to quickly scale in. To the consumer, the capabilities available for provisioning often appear to be unlimited and can be purchased in any quantity at any time. Measured Service. Pay what you drink; Cloud systems automatically control and optimize resource use by leveraging a metering capability at some level of abstraction appropriate to the type of service (e.g., storage, processing, bandwidth, and active user accounts). Resource usage can be monitored, controlled, and reported providing transparency for both the provider and consumer of the utilized service. The cloud can be delivered in different service models (SaaS, PaaS, and IaaS). Deployed in private cloud (dedicated, public shared, or hybrid – mix between both).
Multiple definitions of cloud – pulled from two places:Wikipedia and National Institute of Standards and Technology (NIST) Definition of Cloud Computing: Internet based computing, shared resources, delivered on demand like a public utility. Those shared resources can be networks, servers, storage, applications and services. Over the internent requires less model. Essential Characteristics:On-demand self-service. A consumer can unilaterally provision computing capabilities, such as server time and network storage, as needed automatically without requiring human interaction with each service’s provider. Broad network access. Capabilities are available over the network and accessed through standard mechanisms that promote use by heterogeneous thin or thick client platforms (e.g., mobile phones, laptops, and PDAs). Resource pooling. The provider’s computing resources are pooled to serve multiple consumers using a multi-tenant model, with different physical and virtual resources dynamically assigned and reassigned according to consumer demand. There is a sense of location independence in that the customer generally has no control or knowledge over the exact location of the provided resources but may be able to specify location at a higher level of abstraction (e.g., country, state, or datacenter). Examples of resources include storage, processing, memory, network bandwidth, and virtual machines. Rapid elasticity (dynamic capcity) . Capabilities can be rapidly and elastically provisioned, in some cases automatically, to quickly scale out and rapidly released to quickly scale in. To the consumer, the capabilities available for provisioning often appear to be unlimited and can be purchased in any quantity at any time. Measured Service. Pay what you drink; Cloud systems automatically control and optimize resource use by leveraging a metering capability at some level of abstraction appropriate to the type of service (e.g., storage, processing, bandwidth, and active user accounts). Resource usage can be monitored, controlled, and reported providing transparency for both the provider and consumer of the utilized service. The cloud can be delivered in different service models (SaaS, PaaS, and IaaS). Deployed in private cloud (dedicated, public shared, or hybrid – mix between both).
Here is the family of engineered systems and appliances. On the left you see the purpose build systems, with Exadata designed to run the Oracle database, and Exalogic designed to run Oracle middleware. As well as Exalytics and the database appliances. On the right is the general purpose SPARC SuperCluster which runs the Oracle parallel database, the Oracle fusion middleware parallel java system, and all your enterprise applications. The SPARC SuperCluster runs the Solaris operating system. And that’s extremely important, because it gives our SPARC customers a very, very smooth upgrade path. But it also gives our customers running older hardware from HP and IBM a powerful system to deploy Oracle applications for lower cost and with much better performance, and a management system for the entire stack. Exadata: OLTP/Data Warehouse, Database Consolidation, Database UpgradeExalogic: Middleware Upgrade, Mainframe modernization, Middleware platforms standardizationSPARC SuperCluster: general purpose, multi-tier applications, datacenter consolidation and virtualization
The Oracle SPARC SuperCluster T4-4 is a general purpose engineered solution for running a wide range of enterprise applications with the highest levels of performance and mission critical reliability. The SPARC SuperCluster T4-4 utilizes high performance software from Oracle Exadata and Oracle Exalogic combined with new SPARC T4-4 servers, Oracle Exadata Storage Servers, ZFS Storage Appliance, and InfiniBand technology, and Oracle Solaris 11. The full capabilities of Oracle Exadata along with Exadata Storage Servers will be included with every SPARC SuperCluster T4-4. Exadata Storage Servers are specially optimized for Oracle Database 11gR2 operations and will provide outstanding performance for both transaction-based and decision support workloads. An optional external storage rack can be configured for additional performance and capacity.The SPARC SuperCluster T4-4 provides the accelerated middleware processing of the Oracle Exalogic Elastic Cloud. The new SPARC T4 processor delivers a 5x improvement in single thread performance over the previous generation along with on-chip crypto acceleration providing enhanced data security and privacy without the performance penalty Shared Storage: Includes InfiniBand-attached ZFS storage cluster for high performance, fully redundant disk storage Oracle Enterprise Manager Ops Center is available to manage the software environment on the SPARC SuperCluster. Using Enterprise Manager Ops Center, administrators can perform proactive monitoring and detailed configuration analysis of their SPARC SuperCluster. Supports Oracle Solaris 11 and 10
Multiple definitions of cloud – pulled from two places:Wikipedia and National Institute of Standards and Technology (NIST) Definition of Cloud Computing: Internet based computing, shared resources, delivered on demand like a public utility. Those shared resources can be networks, servers, storage, applications and services. Over the internent requires less model. Essential Characteristics:On-demand self-service. A consumer can unilaterally provision computing capabilities, such as server time and network storage, as needed automatically without requiring human interaction with each service’s provider. Broad network access. Capabilities are available over the network and accessed through standard mechanisms that promote use by heterogeneous thin or thick client platforms (e.g., mobile phones, laptops, and PDAs). Resource pooling. The provider’s computing resources are pooled to serve multiple consumers using a multi-tenant model, with different physical and virtual resources dynamically assigned and reassigned according to consumer demand. There is a sense of location independence in that the customer generally has no control or knowledge over the exact location of the provided resources but may be able to specify location at a higher level of abstraction (e.g., country, state, or datacenter). Examples of resources include storage, processing, memory, network bandwidth, and virtual machines. Rapid elasticity (dynamic capcity) . Capabilities can be rapidly and elastically provisioned, in some cases automatically, to quickly scale out and rapidly released to quickly scale in. To the consumer, the capabilities available for provisioning often appear to be unlimited and can be purchased in any quantity at any time. Measured Service. Pay what you drink; Cloud systems automatically control and optimize resource use by leveraging a metering capability at some level of abstraction appropriate to the type of service (e.g., storage, processing, bandwidth, and active user accounts). Resource usage can be monitored, controlled, and reported providing transparency for both the provider and consumer of the utilized service. The cloud can be delivered in different service models (SaaS, PaaS, and IaaS). Deployed in private cloud (dedicated, public shared, or hybrid – mix between both).
Benchmark disclaimer and data are included below. Based on published results on SPECJEnt, SPARC T4-4 achieved a 2.4x performance advantage over IBM’s published Power 780 benchmark. This reflects roughly 7x better price/performance for Java and demonstrates the multitier capability of SPARC T4 as this benchmark also includes database level operations. And the platform for achieving these results is less than half the cost of the IBM Power 780 configuration used in the benchmark.SPEC and the benchmark name SPECjEnterprise are registered trademarks of the Standard Performance Evaluation Corporation. Other names may be trademarks of their respective owners. Competitive data obtained from http://www.spec.org as of the date located next to the respective claim and this report. See the Website for latest results. SPARC T4-4 cluster: 40,104.86 SPECjEnterprise2010 EjOPS; 1,671 SPECjEnterprise2010 EjOPS performance per processor across the configuration. IBM Power 780 and IBM Power 750 Express: 16,646.34 SPECjEnterprise2010 EjOPS; 1,387 SPECjEnterprise2010 EjOPS performance per processor across the configuration. SPECjEnterprise2010 models contemporary Java-based applications that run on large Java EE (Java Enterprise Edition) servers, backed by network infrastructure and database servers. The Application tier cost of acquisition for four SPARC T4-4 servers with Solaris 10 is $ $467,856 or $11.67/SPECjEnterprise2010 EjOPS. Oracle pricing from https://shop.oracle.com/ on 9/26/1011. The Application tier cost of acquisition for IBM Power 780 (3.86GHz Power7, 512GB RAM, AIX 7.1) is $1,297,956 or $77.97/SPECjEnterprise2010 EjOPS. IBM system pricing is from http://tpc.org/results/FDR/TPCH/TPC-H_1TB_IBM780_Sybase-FDR.pdf, adjusted to license 64 cores (w/o TurboCore). AIX 7.1 pricing is from http://www-304.ibm.com/easyaccess3/fileserve?contentid=214347. $77.97/$11.67=6.7x. Oracle app. tier configuration occupies 20RU of space, 40,104.86/20=2005 SPECjEnterprise2010 EjOPS/RU. IBM app. tier configuration occupies 16RU of space, 16,646.34/16=1040 SPECjEnterprise2010 EjOPS/RU. 2007/1040=1.92x round nearest 2x.
SPARC T4 also generated a solid result in the Data Warehousing space, reflected by excellent performance and price/performance on the TPC-H benchmark. Based on results, our T4 configuration was cheaper and more performant that both IBM and HPs published results. The HP server had 64 cores and the Power7 server had 32 cores. The SPARC T4-4 had 32 cores therefore the SPARC T4-4 server had better total system performance and better per core performance Furthermore, the SPARC T4 processor has a 0.50 core factor and the Power7 and Itanium processors have 1.0 core factors. Thus, the SPARC T4-4 offering is better in performance and price/performance. Substantiation is included below.TPC-H, QphH, $/QphH are trademarks of Transaction Processing Performance Council (TPC). For more information, see www.tpc.org. SPARC T4-4 201,487 QphH@1000GB, $4.60/QphH@1000GB, avail 10/30/2011, 4 processors, 32 cores, 256 threads; SPARC Enterprise M8000 209,533.6 QphH@1000GB, $9.53/QphH@1000GB, avail 09/22/11, 16 processors, 64 cores, 128 threads; IBM Power 780 QphH@1000GB, 164,747.2 QphH@1000GB, $6.85/QphH@1000GB, avail 03/31/11, 8 processors, 32 cores, 128 threads; HP Integrity Superdome 2 140,181.1 QphH@1000GB, $12.15/QphH@1000GB avail 10/20/10, 16 processors, 64, cores, 64 threads.http://www.tpc.org/results/individual_results/Oracle/Oracle_T4-4_1TB_TPCH_ES_092611.pdf
Multiple definitions of cloud – pulled from two places:Wikipedia and National Institute of Standards and Technology (NIST) Definition of Cloud Computing: Internet based computing, shared resources, delivered on demand like a public utility. Those shared resources can be networks, servers, storage, applications and services. Over the internent requires less model. Essential Characteristics:On-demand self-service. A consumer can unilaterally provision computing capabilities, such as server time and network storage, as needed automatically without requiring human interaction with each service’s provider. Broad network access. Capabilities are available over the network and accessed through standard mechanisms that promote use by heterogeneous thin or thick client platforms (e.g., mobile phones, laptops, and PDAs). Resource pooling. The provider’s computing resources are pooled to serve multiple consumers using a multi-tenant model, with different physical and virtual resources dynamically assigned and reassigned according to consumer demand. There is a sense of location independence in that the customer generally has no control or knowledge over the exact location of the provided resources but may be able to specify location at a higher level of abstraction (e.g., country, state, or datacenter). Examples of resources include storage, processing, memory, network bandwidth, and virtual machines. Rapid elasticity (dynamic capcity) . Capabilities can be rapidly and elastically provisioned, in some cases automatically, to quickly scale out and rapidly released to quickly scale in. To the consumer, the capabilities available for provisioning often appear to be unlimited and can be purchased in any quantity at any time. Measured Service. Pay what you drink; Cloud systems automatically control and optimize resource use by leveraging a metering capability at some level of abstraction appropriate to the type of service (e.g., storage, processing, bandwidth, and active user accounts). Resource usage can be monitored, controlled, and reported providing transparency for both the provider and consumer of the utilized service. The cloud can be delivered in different service models (SaaS, PaaS, and IaaS). Deployed in private cloud (dedicated, public shared, or hybrid – mix between both).
The architecture of the Exadata storage cells is that instead of a server asking for a block of data and the block of data coming back to the server, we’re actually constructing queries that are executed in parallel across the different storage devices. Flash memory in the storage device caches the intermediate results, as well as do advanced compression and allow you to run queries and do data warehousing against compressed and encrypted data. This further demonstrates the value of integrated hardware and software. It’s not possible for a conventional storage system to equal this. And so there is no possible way from any other vendor that you can run Oracle Database 11g faster than it runs on Exadata or on SuperCluster by virtue of the fact that we have an architecturally different coupling between the database and the storage. This result of engineering the hardware and software together between the database and storage is very fundamental because that’s what gives us the IO capability and performance of SuperCluster, as well as Exadata.
Multiple definitions of cloud – pulled from two places:Wikipedia and National Institute of Standards and Technology (NIST) Definition of Cloud Computing: Internet based computing, shared resources, delivered on demand like a public utility. Those shared resources can be networks, servers, storage, applications and services. Over the internent requires less model. Essential Characteristics:On-demand self-service. A consumer can unilaterally provision computing capabilities, such as server time and network storage, as needed automatically without requiring human interaction with each service’s provider. Broad network access. Capabilities are available over the network and accessed through standard mechanisms that promote use by heterogeneous thin or thick client platforms (e.g., mobile phones, laptops, and PDAs). Resource pooling. The provider’s computing resources are pooled to serve multiple consumers using a multi-tenant model, with different physical and virtual resources dynamically assigned and reassigned according to consumer demand. There is a sense of location independence in that the customer generally has no control or knowledge over the exact location of the provided resources but may be able to specify location at a higher level of abstraction (e.g., country, state, or datacenter). Examples of resources include storage, processing, memory, network bandwidth, and virtual machines. Rapid elasticity (dynamic capcity) . Capabilities can be rapidly and elastically provisioned, in some cases automatically, to quickly scale out and rapidly released to quickly scale in. To the consumer, the capabilities available for provisioning often appear to be unlimited and can be purchased in any quantity at any time. Measured Service. Pay what you drink; Cloud systems automatically control and optimize resource use by leveraging a metering capability at some level of abstraction appropriate to the type of service (e.g., storage, processing, bandwidth, and active user accounts). Resource usage can be monitored, controlled, and reported providing transparency for both the provider and consumer of the utilized service. The cloud can be delivered in different service models (SaaS, PaaS, and IaaS). Deployed in private cloud (dedicated, public shared, or hybrid – mix between both).
So what we exactly we mean by a complete, integrated and business-driven cloud management solution?There are three key aspects of Oracle Enterprise Manager 12c that help accomplish this: Complete Cloud Lifecycle Solution To start with, Oracle Enterprise Manager 12c contains solutions to manage all phases of the building,managing, and consuming an enterprise cloud. Using Oracle Enterprise manager 12c you can build and manage a rich catalog of cloud services – whether it is Infrastructure-as-a-Service, Database-as-a-Service, or Platform-as-a-Service , all from a single product. Integrated Cloud Stack ManagementSecondly, Oracle Enterprise Manager 12c enables integrated management of the entire cloud stack – all the way from application to disk. Oracle Enterprise Manager 12c therefore eliminates much of the integration pains and costs that customers would have to otherwise incur by trying to create a cloud environment by integrating multiple point solutions. Business-Driven Clouds Finally, Oracle Enterprise Manager 12c enables creation of application-aware and business-driven clouds that has deep insight into applications,business services and transactions. Applications – whether they are packaged or home grown – power your businesses and therefore it is critical that an enterprise Cloud platform is not only able to run these applications but also has deep business insight and visibility. As the leading providers of business applications and the middleware that many of your custom applications are built on top of, we are able to offer you a cloud solution that is optimized for business services.
Multiple definitions of cloud – pulled from two places:Wikipedia and National Institute of Standards and Technology (NIST) Definition of Cloud Computing: Internet based computing, shared resources, delivered on demand like a public utility. Those shared resources can be networks, servers, storage, applications and services. Over the internent requires less model. Essential Characteristics:On-demand self-service. A consumer can unilaterally provision computing capabilities, such as server time and network storage, as needed automatically without requiring human interaction with each service’s provider. Broad network access. Capabilities are available over the network and accessed through standard mechanisms that promote use by heterogeneous thin or thick client platforms (e.g., mobile phones, laptops, and PDAs). Resource pooling. The provider’s computing resources are pooled to serve multiple consumers using a multi-tenant model, with different physical and virtual resources dynamically assigned and reassigned according to consumer demand. There is a sense of location independence in that the customer generally has no control or knowledge over the exact location of the provided resources but may be able to specify location at a higher level of abstraction (e.g., country, state, or datacenter). Examples of resources include storage, processing, memory, network bandwidth, and virtual machines. Rapid elasticity (dynamic capcity) . Capabilities can be rapidly and elastically provisioned, in some cases automatically, to quickly scale out and rapidly released to quickly scale in. To the consumer, the capabilities available for provisioning often appear to be unlimited and can be purchased in any quantity at any time. Measured Service. Pay what you drink; Cloud systems automatically control and optimize resource use by leveraging a metering capability at some level of abstraction appropriate to the type of service (e.g., storage, processing, bandwidth, and active user accounts). Resource usage can be monitored, controlled, and reported providing transparency for both the provider and consumer of the utilized service. The cloud can be delivered in different service models (SaaS, PaaS, and IaaS). Deployed in private cloud (dedicated, public shared, or hybrid – mix between both).
It is important to understand how the SSC can be virtualized to understand its use as a consolidation platform. Using Oracle VM for SPARC (OVM SPARC) the T4-4 compute nodes can be partitioned into three types of partitions. One is a Database Domain that is used to run S11 and Oracle 11gR2. There can be only one DB Domain per T4-4. A DB Domain does not have to include all processors in a T4-4 node. There is a maximum of 3 domains on a T4-4 node that has a DB domain. DB Domains will access the Exadata storage cells. Only DB Domains running Oracle 11gR2 can use the Exadata storage cells. General Purpose (GP) domains can run S10 or S11 and has their name implies are used to run general purpose applications such as Oracle Ebiz, Siebel, PSFT or even third-party applications. GP Domains will use the ZFS Storage appliance for its disk requirements. GP domains can also access legacy SAN’s. Exalogic domains are domains that run the Exalogic Elastic Cloud software. Exalogic domains will use the ZFSSA for storage. Up to 4 domains can be deployed on any T4-4 nodes but since only one DB Domain is possible on a T4-4 node the other three domains would have to be GP Domains or Exalogic Domains. This slide shows all three types of domains indicating how to virtualize/partition an SSC T4-4. It also shows how it is possible to consolidate app tier and db tier on a single SSC T4-4. It shows that all four T4-4 nodes are used for database domains and are clustered using RAC. These domains will have exclusive access to the 6 Exadata Storage nodes. The other domains nodes are partitioned into 3 GP domains and 1 exalogic domain. Note that only GP domains can run S10. If a customer wants a T4-4 node to run just an S10 GP Domain then they still need to set up a small S11 control domain as it is not allowed to have a node running just S10.Deploying DB’s in containers in DB Domains is not supported.Lastly, live migration is not supported on SSC T4-4.
With SPARC SuperCluster as a foundation for the enterprise cloud, one of the big questions, obviously, is how do I take advantage of this? You put together storage, and you put together networking and you put together servers, and that’s different than what I have today. But one of the things that Oracle has worked on here is making it seamlessly integrate into existing data centers. So you have your existing data …. You can connect to your existing storage infrastructure and your existing data. Database 11 is, obviously, accelerated by ASM in the Exadata storage cells. But you can easily and seamlessly connect to your existing SAN and run you existing data. You can start out on a SuperCluster half rack and you can move existing applications. You can run them side-by-side with Database 11. You could ignore Database 11 in a virtualized environment and use them with your existing environment. You can approach at your own pace, upgrading existing applications and using them in a virtualized environment starting out with a relatively small system.Up to 8 SuperClusters can be clustered together to create a very large system with up to 32 T4-4 servers, 48 Exadata storage nodes and 8 ZFS Storage appliances. Note that it is possible to attach Exadata expansion racks to SuperClusters to increase storage capacity without increasing compute capacity. If Exadata expansion racks are used then that reduces the number of SuperCluster racks that can be clustered. The total number of SuperCluster and storage racks that can be combined are 8 but can be a mix.
Multiple definitions of cloud – pulled from two places:Wikipedia and National Institute of Standards and Technology (NIST) Definition of Cloud Computing: Internet based computing, shared resources, delivered on demand like a public utility. Those shared resources can be networks, servers, storage, applications and services. Over the internent requires less model. Essential Characteristics:On-demand self-service. A consumer can unilaterally provision computing capabilities, such as server time and network storage, as needed automatically without requiring human interaction with each service’s provider. Broad network access. Capabilities are available over the network and accessed through standard mechanisms that promote use by heterogeneous thin or thick client platforms (e.g., mobile phones, laptops, and PDAs). Resource pooling. The provider’s computing resources are pooled to serve multiple consumers using a multi-tenant model, with different physical and virtual resources dynamically assigned and reassigned according to consumer demand. There is a sense of location independence in that the customer generally has no control or knowledge over the exact location of the provided resources but may be able to specify location at a higher level of abstraction (e.g., country, state, or datacenter). Examples of resources include storage, processing, memory, network bandwidth, and virtual machines. Rapid elasticity (dynamic capcity) . Capabilities can be rapidly and elastically provisioned, in some cases automatically, to quickly scale out and rapidly released to quickly scale in. To the consumer, the capabilities available for provisioning often appear to be unlimited and can be purchased in any quantity at any time. Measured Service. Pay what you drink; Cloud systems automatically control and optimize resource use by leveraging a metering capability at some level of abstraction appropriate to the type of service (e.g., storage, processing, bandwidth, and active user accounts). Resource usage can be monitored, controlled, and reported providing transparency for both the provider and consumer of the utilized service. The cloud can be delivered in different service models (SaaS, PaaS, and IaaS). Deployed in private cloud (dedicated, public shared, or hybrid – mix between both).
In this comparison 2 x E25Ks with 144 processors and 288 cores are being replaced by one SPARC SuperCluster half rack with 64 cores. The software licensed for these systems based on # of cores and multipliers includes Oracle Database Enterprise Edition. The majority of the total cost of ownership over 5 years is due to software licensing and software service. Assumptions:Oracle software licenses are under a contract which allows the licenses to be transferred on a core for core basis.Requirement to purchase Exadata Storage software license with SuperCluster acquisition.A 60% discount is assumed to be applied to the software service. Hardware and hardware service have no discount assumptions applied. 50% of E25K and SuperCluster resources allocated to database, therefore, 50% of total Oracle Database Enterprise Edition licensed on a per core basis.It is recommended that account teams generate customer specific scenarios for legacy SPARC upgrades to SPARC SuperCluster based on customer specific assumptions. Example Value Navigator cases for SuperCluster are available in the library.Value Navigator: http://tools-insight.oraclecorp.com/sbat/faces/Main.jspx?_afrLoop=84219636575602214&_afrWindowMode=0&_adf.ctrl-state=j156n84bn_30SPARC SuperCluster Sizing Guide:http://my.oracle.com/content/groups/public/@empl/@pd/@sss/@servers/@sparc/documents/webcontent/cnt1354301.pdfThis TCO example may be used externally.
This provides another SPARC consolidation scenario of 10 x M5000s to a SSC half rack. It is recommended that account teams generate customer specific scenarios for legacy SPARC upgrades to SPARC SuperCluster based on customer specific assumptions. Example Value Navigator cases for SuperCluster are available in the library.Assumptions:Oracle software licenses are under a contract which allows the licenses to be transferred on a core for core basis.Requirement to purchase Exadata Storage software license with SuperCluster acquisition.A 60% discount is assumed to be applied to the software service. Hardware and hardware service have no discount assumptions applied. 50% of resources allocated to database, therefore, 50% of total cores Oracle Database Enterprise Edition licensed on a per core basis.50% of resources allocated to applications, therefore, 50% of the total CPUs licensed for WebLogic Suite.Value Navigator: http://tools-insight.oraclecorp.com/sbat/faces/Main.jspx?_afrLoop=84219636575602214&_afrWindowMode=0&_adf.ctrl-state=j156n84bn_30SPARC SuperCluster Sizing Guide:http://my.oracle.com/content/groups/public/@empl/@pd/@sss/@servers/@sparc/documents/webcontent/cnt1354301.pdfThis TCO example may be used externally.
Substantiation and Assumptions:In order to provide a valid relative performance comparison, IBM TPC-H benchmark result was taken as a baseline, and the following adjustments were made:1) Assumed a 63% perf improvement for the p780 going from TurboCore (4.14GHz) to using all cores at 3.92GHz, per IBM’s rPerf2) SSC performance extrapolated from published TPC-H result 1xT4-4 server3) Used an Oracle DB stack for IBM, instead of Sybase IQ (which was used in their TPC-H benchmark submission)4) Priced AIX EE and PowerVM on IBM configuration, to match features offered at no additional cost with Solaris and Ops CenterIBM benchmark: http://tpc.org/results/FDR/TPCH/TPC-H_1TB_IBM780_Sybase-FDR.pdfIBM 780 Software license price calculation:Software is licensed on all of the cores. Per-cpu list prices:- Oracle EE: $47.5k- Oracle RAC: $23k- Advanced Partitioning: $11.5kTotal: $82k per cpuAt 60% Discount: $32.8k per cpuCore multiplier for Power7: 1.0IBM 780 Software service price calculation:$4.2M * 0.22 * 5 = $4.6MSuperCluster Software license calculation:Software is licensed on all of the cores. Per-cpu list prices:- Oracle EE: $47.5k- Oracle RAC: $23k- Advanced Partitioning: $11.5kTotal: $82k per cpuAt 60% Discount: $32.8k per cpuCore multiplier for T4: 0.5Exadata Storage: $10k per disk ($4K at 60% discount)128 cores x 0.5 multiplier x $32.8 = $2.1M 72 Exadata disks x $4k = $288kSuperCluster Software service calculation:$2.4M * 0.22 * 5 = $2.64MIBM system and service pricing publicly available at the following sites: http://www-01.ibm.com/common/ssi/cgi-bin/ssialias?infotype=an&subtype=ca&htmlfid=897/ENUS111-201&appname=isource&language=enus http://www-01.ibm.com/common/ssi/rep_ca/1/897/ENUS111-201/ENUS-111-201-List_prices_2011_10_12.PDFAssumptions:No transfer of licensing, new deployment.60% discount applied to software.List price applied to hardware.
Assumptions:Oracle software licenses are under a contract which allows the licenses to be transferred on a core for core basis.Requirement to purchase Exadata Storage software license with SuperCluster acquisition.A 60% discount is assumed to be applied to the software service. Hardware and hardware service have no discount assumptions applied. 100% of resources allocated to database, therefore all cores licensed.
Multiple definitions of cloud – pulled from two places:Wikipedia and National Institute of Standards and Technology (NIST) Definition of Cloud Computing: Internet based computing, shared resources, delivered on demand like a public utility. Those shared resources can be networks, servers, storage, applications and services. Over the internent requires less model. Essential Characteristics:On-demand self-service. A consumer can unilaterally provision computing capabilities, such as server time and network storage, as needed automatically without requiring human interaction with each service’s provider. Broad network access. Capabilities are available over the network and accessed through standard mechanisms that promote use by heterogeneous thin or thick client platforms (e.g., mobile phones, laptops, and PDAs). Resource pooling. The provider’s computing resources are pooled to serve multiple consumers using a multi-tenant model, with different physical and virtual resources dynamically assigned and reassigned according to consumer demand. There is a sense of location independence in that the customer generally has no control or knowledge over the exact location of the provided resources but may be able to specify location at a higher level of abstraction (e.g., country, state, or datacenter). Examples of resources include storage, processing, memory, network bandwidth, and virtual machines. Rapid elasticity (dynamic capcity) . Capabilities can be rapidly and elastically provisioned, in some cases automatically, to quickly scale out and rapidly released to quickly scale in. To the consumer, the capabilities available for provisioning often appear to be unlimited and can be purchased in any quantity at any time. Measured Service. Pay what you drink; Cloud systems automatically control and optimize resource use by leveraging a metering capability at some level of abstraction appropriate to the type of service (e.g., storage, processing, bandwidth, and active user accounts). Resource usage can be monitored, controlled, and reported providing transparency for both the provider and consumer of the utilized service. The cloud can be delivered in different service models (SaaS, PaaS, and IaaS). Deployed in private cloud (dedicated, public shared, or hybrid – mix between both).