Oracle databáze – Konsolidovaná Data Management Platforma

1
Consolidate &
Modernize Database
Jaroslav Malina, Josef Krejčí, Josef Šlahůnek
Oracle Czech
Copyright © 2020, Oracle and/or its affiliates. All rights reserved. |

Safe harbor statement
The following is intended to outline our general product direction. It is intended for information purposes only,
and may not be incorporated into any contract. It is not a commitment to deliver any material, code, or
functionality, and should not be relied upon in making purchasing decisions. The development, release,
timing, and pricing of any features or functionality described for Oracle’s products may change and remains at the
sole discretion of Oracle Corporation.

Data Management
Market

Confidential – Oracle Internal and Approved OPN Partners Only

Copyright © 2020, Oracle and/or its affiliates. All rights reserved.
7
DATA – The world’s most valuable resource like Water
DATA
Unique
Data cannot be substituted by other
data, because each carries different
information.
[Less Transformation is Better]
Always There
Data can be consumed over and over
again by different parties.
[Easy and Repeatable Access is a Must]
Experience
Data can be accurately evaluated only
after data have been experienced.
[Time to Experience is Important]
THE
OPPORTUNITY
$430B
advantage for
data driven
organizations
- IDC
19X
likely to be
more profitable
- McKinsey
10%
increase in data
accessibility
translates into
$65.7M in net
income
- Baseline

And We need to prepare for the
IoT/5G Data Tsunami That is Arriving
Petabytes/DAY
Smart
City
Petabytes/DAY
Connected
Factories
Petabytes/DAY
Smart
Hospital
Petabytes
/DAY/CAR
Self-Driving

Data Strategy
Will be Critical
to a Successful Digital Transformation
And Surviving todays Challenges

Why
Consolidate?

Complexity of the IT infrastructure
Leads to…Data Integrity, Security, Availability Challenges
AND high integration and Management costs …
Business Drivers for Database Consolidation
Custom Apps
Legacy Apps
E-Business Suite PeopleSoft
JD Edwards
SAP/Other
Data warehouse
Analytics &
reporting
On-Premises Enterprise applications on disparate locations
Many copies
and
instances Test / Dev Data Marts HA / DR
Machine
Learning
Blockchain
Cloud
{ }
XML/JSON
IoT
Spatial &
Graph
Hyperscale
Persistent
Memory
Microservices
Events
File
OLTP Analytic
{rest
}
REST
New Innovations
New Data Types
New Workloads
Data Fragmentation across multiple data stores
Leads to …High Operational costs, Data Complexity …

Single purpose databases are advocated by cloud vendors and startups
Combining “best-of-breed” drives “worst-of-weaknesses”
• Use of single-purpose databases has been the source
of a large increase in data management complexity
• Require apps to use separate proprietary APIs for each
database
• Separately implement security policies for every
database technology
• Unique high availability and scalability tools for each
database type
• Complex management and integration
• End-to-End security, availability, scalability,
consistency, etc. means weakest of the databases that
are used
AWS and other vendors want customers to run multiple
Single-Purpose databases because it drives up
subscriptions & services (= More Revenue $$$$)
Amazon
Aurora
Amazon DocumentDB
Amazon DynamoDB
Amazon
Neptune
Amazon Quantum
Ledger Database
Amazon RedShift
Amazon
Timestream
Amazon
ElastiCache
Data Warehouse
Document and NoSQL
Transactions And other single
purpose databases!

• Modern applications require many different:
– Data types - Relational, Document, Spatial, Graph, etc.
– Workloads - Transactions, analytics, ML, IoT, etc.
• Each data type and workload requires different database algorithms
• Two possible data strategies:
– Use single-purpose “best-of-breed” database for each data type and
workload
– Use a converged database for all data types and workloads
Oracle’s Strategy:
– Run Converged, Open, Oracle Database for multiple data types
and workloads
To become data-driven customers need a modern data strategy
Many specialized databases vs one converged database

Oracle solution
16 Copyright © 2022, Oracle and/or its affiliates

Two Contrasting Database Architecture Strategies
Copyright © 2019, Oracle and/or its affiliates. All rights reed.
Spatial
ML
Blockchain
Document
Graph
Reporting
Single-purpose database for each data type
and workload
A Converged database that supports
multiple data types and workloads

Converged database to dramatically reduce complexity and cost
Oracle Database Supports All Data Management Needs
RDBMS Document
Graph Time Series
Key Value Spatial
Benefits
Less
Integration
Lower
Costs
$
Higher
Performance
Consistent
Security
Rapid
Development
Machine
Learning
Higher
Consolidation
Simpler
Management
Higher
Availability
Greater
Scale
Zero Data Loss
Protection

Roadmap to Database Consolidation
Reduce data fragmentation
with a converged database
architecture
Consolidate databases &
infrastructure
Optimize mission critical
workloads
Deploy it most advanced
database platform –
Oracle Exadata
Run high-performance
database workloads in the
Cloud
Preserve existing
investments
As-Is
IT complexity
Data Fragmentation
Avoidable operational
expenses
Standardization
1 Consolidation
2 Transform in the cloud
3

Exadata – The perfect platform for
database consolidation
Effective Consolidation with
Oracle Multitenant
Oracle Database - The most
advanced database platform
Three Pillars of Database Consolidation
Marketing
Manufacturing
Service
IT
Finance
R&D
Sales
HR
• Minimize CAPEX, more
applications per server
• Minimize OPEX by managing
many as one
• Increase agility , provisioning,
relocation , cloning
• High Consolidation density. For
ALL database workloads
• 69% more efficient IT staff
• High 5-year ROI
• Deliver superior performance,
availability and security
• Improve IT operations.
Accelerate development
• Unlock new value from ANY
data

Oracle Multi Tenant on Exadata
Scaling-Out Database Containers keeping Transactional Consistency
Application Containers
Scaling-Out
Stateless Applications
(Twelve-Factor like apps)
Database Containers
Scaling-Out
Stateful Applications
(Databases)
Keeping
Transactional Consistency
Server
Server
Server
Server
Server
Server
Server
Kubernetes Cluster
Docker Docker Docker
Storage
Server
Disk
Flash
Storage
Server
Disk
Flash
Storage
Server
Disk
Flash
Storage
Server
Disk
Flash
Storage
Server
Disk
Flash
Storage
Server
Disk
Flash
RAC
Node
RAC
Node
RAC
Node
RAC
Node
RAC
Node
RAC
Node
Database Cluster Container
PDB
PDB
Exadata Scale-Out Grid
PDB
Docker
PDB
CPU
RAM
Disk
CPU
RAM
Disk
IOPS
CPU
Disk
RAM
IOPS
CPU
Disk
RAM

Oracle Multitenant implementations in The Czech Republic
Tenant 1 Tenant 2 Tenant 3
Master
Sberbank Czech Republic
MT on Oracle Sparc (2019)
Benefit: Saving: RAM
Prazska energreticka - utility distributor
MT on HPx86 and Linux (2021)
DB migration project from 10g, 11c, 12c to 19c started in April 21
Benefit: B 11c, 12c to 19c migration / from schema consolidation to PDB consolidation
J&T bank - private investment bank
MT for Dev/test environment on Exadata, (2018)
Average 50 PDB per container DBs
Benefits: Saving: RAM, CPU, Oracle DB License
• Saving provisioning and patching time
ČSOB stavební spořitelna
MT for Dev/test environment on Exadata, (2019)
Average 20 PDB per container DBs
Benefits: Saving: RAM, CPU, Oracle DB License
• Saving provisioning time
• DB 11c, 12c to 19c migration

Exadata is the Premier Platform for Database Consolidation
Consolidate your databases, data types, and workloads into one database
• Optimized Database System – scale-out, database
optimized compute, networking, and storage for fastest
performance and lowest cost
• Smart System Software – specialized algorithms vastly
improve OLTP, Analytics, Consolidation
• Automated Management – fully automated and
optimized configuration, performance, fault-tolerance,
updates
• Multitenant Architecture - Self-contained pluggable
databases in a shared Container Database
Reduce Cost
Simplicity
Availability
Performance
Security

Oracle Exadata is The Best Platform for Oracle Database
Exadata delivers unmatched Oracle Database capabilities for
60 Exadata-only features for Oracle Database
Scalability
1. Infiniband Cluster Interconnect
2. Remote Direct Memory Access on Storage I/O
3. Exadata High Redundancy Storage
4. High Performance I/O – 6.57 million IOPS from
SQL
5. Low Latency I/O – 200 microseconds
6. In-Memory Data Mirroring
7. Hybrid Columnar Compression
8. Bloom Filter Joins
9. In-Memory Columnar Tables
10. In-Memory External Tables
11. Memory Optimized Key/Value Data
12. Higher Consolidation Density
Performance
1. Active/Active IB Network
2. Exadata Smart Write Back, Smart Flash Logging,
Smart Scan, and Reverse Offload
3. Fastest Redo Apply and Instance Recovery
4. Efficient re-silver rebalance after Flash failure
5. I/O latency capping for reads and writes
6. Cell IO timeout threshold
7. Smart Write Back Flash Cache Persistence
8. I/O and network resource management
9. Cell to cell offload for disk repair
10. Cell-to-Cell Rebalance Preserves FlashCache
11. Appliance Mode Support
Security
1. Full Stack Patching
2. Minimal Attack Surface
3. Pre-Scanned & fixed system stack using STIG,
Nessus, and Qualys
4. Advanced Intrusion Detection Environment
(AIDE) –similar to virus scanners
5. SGX Integration in Exadata Storage Cells
Availability
1. Fast Node and Cell Death Detection
2. Fast Network Failure Detection
3. Redundancy Protection on cellsrv shutdown
4. Redundancy Protection on Cell shutdown
5. Reduced Brownout for Instance Recovery
6. ILOM Hang Detection and Repair
7. Automatic ASM Mirror Read on IO error corruption
8. IO error prevention with Exadata disk
scrubbing/ASM Corruption repair
9. Corruption Prevention with HARD support
10. Elimination of false-positive drive failures
11. Redundancy Check During Power Down
12. Blue OK-to-remove LED Light notification
13. Health Factor on predictively failed disks
14. Disk Confinement
15. I/O hang detection and repair
16. Drop Hard Disk for Replacement
17. Drop BBU for Replacement
Efficiency
1. Exadata Elastic Configuration
2. Cell Alert Summary
3. Flash and Disk Lifecycle Management Alerts
4. Auto Online
5. Auto Disk Management
6. Priority Rebalance Support
7. EM Failure Reporting
8. Failure Monitoring on Database Servers
9. Updating Database nodes with Patchmgr
10. Optimized and Faster Exadata Patching
11. Custom Diagnostic Package for Cell alerts
12. VLAN support and automation
13. Exachk – full stack health check with critical issue alerts
14. Automatic Statistics
15. Automatic Indexing
Availability Security Performance
Scalability Efficiency

Do-it-Yourself 48% Higher Cost than Full
Stack systems
(4-year Financial Analysis Mission Critical IT for Typical $2B Enterprise)
TRADITIONAL IT DATA CENTER
4-YEAR TOTAL
ORACLE EXADATA X8M ON-
PREMISES 4-YEAR TOTAL
4-year
Infrastructure
Costs,
including
Maintenance
and
Operational
Costs
for
System
&
Database.
Includes
Oracle
Database
Software
Licenses.
Oracle Exadata X8M Powers Multi-Cloud Strategy, David Floyer, Wikibon. December 10, 2019
$39.5
$26.8
DIY x86 database solutions
cost more than Exadata
Higher 4-year TCO
48%
2.5x Higher operating costs
“The time for deploying “Do-it-Yourself”
(DiY) systems for any Tier-1 database
supporting larger-scale mission-critical
systems of record is well and truly over”
WIKIBON: Oracle EDX8M Powers MultiCloud Strategy

Level of automation
Your Choice of Deployment model
The best place to run Oracle Database - on-premises and in the cloud
On-Premises
Cloud@Customer
Public Cloud
Customer Data Center
Oracle Data Center
Subscription with Universal Credit-based consumption Purchased
Co-Managed Customer Managed
Cloud@Customer
Cloud Service
Cloud
Fully Oracle Managed Co-Managed
Fully Oracle Managed
On ExaC@C

What is new inside Exadata
27

Innovation Excellence Since 2008
Oracle Exadata – The Game Changer
+ Flash Cache Write-Back
+ 1/8 Rack Release
+ Flash Cache
+ Hybrid Columnar
Compression
+ Infiniband
+ PMEM
+ ROCE
+ 160% Performance
Boost
+ PCIe Gen4
+ Storage Index Persistence
+ Columnar Cache
Persistence
Autonomous
Database
28
SQL Offload
IO Resource
Management
Exadata
V1
Oracle
Database
Machine
2008
Sun Oracle
Database
Machine
Exadata
V2
2009
Extreme
Performance for
OLTP, Analytics,
Consolidation
Exadata
X2
2010
Exadata
X3
2012
In-Memory
Database
Machine
Exadata
X4
2013
+ Elastic Config
+ Columnar Flash Cache
+ VM Support
Best Platform for
OLTP, Analytics,
Consolidation
Exadata Cloud
Service
Exadata
X5
2015
Exadata
Cloud@Customer
Exadata
X6
2016
+ Storage Tier
In-Memory Analytics
+ Hot Swap Flash Card
Exadata
X7
2017
Exadata
X9M
2021
+ Extended Storage
+ Automatic Indexing
+ 60% Performance
Boost
Exadata
X8
2019
Exadata
X8M
2019
Autonomous
Database on
Exadata
Cloud@Customer
Copyright © 2022, Oracle and/or its affiliates

Database Platform Leadership Since 2008
64 64 96 128 192 288 352 384 384 384 512 8 X
256 576 1152 2048 4096 6144 12 TB 12 TB 12 TB 12 TB 16 TB 64 X
0 5.3 5.3 22.4 44.8 89.6 179.2 358 358 358 358 67 X
168 336 504 504 672 1344 1344 1680 2.35 PB 2.35 PB 3 PB 17 X
20 40 40 40 40 40 40 40 40 100 100 5 X
8 24 184 400 400 400 400 800 800 800 800 100 X
14 50 75 100 100 263 301 350 560 560 1050 75 X
.05 1 1.5 1.5 2.66 4.14 5.6 5.97 6.57 16 27.6 552 X
V1 – X9M
Growth
V1
Sep 2008
Xeon E5430
Harpertown
V2
Sep 2009
Xeon E5540
Nehalem
X2
Sep 2010
Xeon X5670
Westmere
X3
Sep 2012
Xeon E5-2690
Sandy Bridge
X4
Nov 2013
E5-2697 v2
Ivy Bridge
X5
Dec 2014
E5-2699 v3
Haswell
X6
Apr 2016
E5-2699 v4
Broadwell
X7
Oct 2017
Xeon 8160
Skylake
X8M
Sep 2019
Xeon 8260
Cascade Lake
X8
Apr 2019
Xeon 8260
Cascade Lake
X9M
Sep 2021
Xeon 8358
Ice Lake

Exadata X9M: Extreme Performance Scale-out Database Platform
Database Server
High-Capacity (HC) Storage
Extreme Flash (EF) Storage
Extended (XT) Storage
30
Compute Capacity
1,216 DB Cores
38 TB Memory
Storage Capacity
576 Storage Cores
27 TB PMem
920 TB NVMe Flash
3.8 PB Raw Disk
Each rack
has up to...
• Scale-Out 2 Socket Database Servers
• Latest 32-core Intel Ice Lake CPUs
• Up to 2TB Memory
• 8-socket database server for SMP workload
• Ultra-fast 100Gb/s RDMA over Converged
Ethernet (RoCE) Internal Fabric
• Scale-Out intelligent 2-Socket Storage Servers
• Latest 16-core Intel Ice Lake CPUs
• 1.5 TB Persistent Memory per storage server
• Three tiers of storage: PMem, NVMe Flash, HDD

• A New silicon technology innovatively implemented in
Exadata
• Capacity, performance, and price are between DRAM and flash
• Reads at memory speed – much faster than flash
• Writes survive power failure unlike DRAM
• Intel® Optane™ Persistent Memory
• Leading implementation integrated into Exadata storage
• Exadata implements sophisticated algorithms to
maintain integrity of data on PMem during failures
• Call special instructions to flush data from CPU cache to PMem
• Complete or backout sequence of writes interrupted by a crash
Persistent Memory (PMem)
31
Persistent Memory
DRAM
FLASH
DISK
Faster
Higher
Cost
per
GB

Processing in Storage Eliminates Network Bottlenecks
• Flash Storage has Become Very Fast
• Storage bandwidth easily outpaces fast networks, such that ...
• Just 4 NVMe Flash drives can saturate a network
• Conventional Systems Must Move Data to Compute
• There is no ability to filter out data in storage, and since ...
• Analytics requests can process immense amounts of data, ...
• The network quickly becomes the bottleneck
• Exadata Moves Processing to Storage
• Exadata storage also runs Oracle Database software, able to ...
• Join, sort and filter out data not needed in the final result, ....
• Sending much less data across the (very fast) internal network, ...
• Allowing the full performance of Flash to be realized
Example: How Exadata Realizes the Full Performance of Flash
32
0
200
400
600
800
1000
Number of flash drives
Throughput
GB/sec
All-Flash
Array
Exadata
96% of flash
performance is lost
to a network bottleneck

Yielding Substantially Higher ROI
• Specialized Software, Hardware, Protocols
• Oracle owns/controls most of the tech stack, allowing for ...
• Purpose-built servers, optimal configurations and
customized algorithms, with ...
• Seamless integration between layers
• Dedicated to Oracle Databases
• Nothing else runs on Exadata, enabling ...
• “Smart”, Oracle-aware processing, and ...
• Automated workload prioritizations
• Across BOTH Compute and Storage
• Data-intensive work is offloaded to “Smart” storage, ...
• Avoiding internal network bottlenecks, and ...
• Utilizing the entire platform simultaneously
Exadata is a Uniquely Powerful Platform
33
Compute
Servers
Storage
Servers
Network
Internal
Oracle Protocols
Oracle Software
Oracle Hardware
Exadata Platform
Concepts

PMEM Accelerator
PCI NVMe Flash
• Unique: Innovative, modern technologies tuned for high-volume,
low-latency random I/O
• 100 Gb/s RDMA over Converged Ethernet 100 (RoCE) network fabric
• Persistent Memory (PMEM) Data and Commit Accelerators
• Scale-out Database & Storage servers
• Automatic data tiering between PMEM, Flash and Disk
• Unique: Elimination of DB cluster coordination bottlenecks
• Direct-to-Wire Protocol = 3x faster inter-node OLTP messaging
• Smart Fusion Block Transfer eliminates inter-node log write
• RDMA protocol coordinates transactions between nodes
• Unique: Instant detection, handling of failed/failing components
• Automatic discovery of server failures without timeout
• Sub-second redirection of I/Os around sick devices
Fastest OLTP
Exadata
Database Cluster
100g RoCE Network
Exadata
Storage Server
PMEM Accelerator
PCI NVMe Flash
PMEM Accelerator
PCI NVMe Flash
34
RDMA

• Unique: Smart Scan (SQL Offload)
• Data-intensive processing* runs in Exadata Storage,
bypassing network bottlenecks and freeing up DB CPUs
• Unique: Tiered Flash Cache
• Active data is automatically cached on PCI NVMe Flash,
inactive data on low cost, high-capacity disks
• Unique: Storage Indexes
• Eliminates I/O not relevant to a particular query
• Unique: Hybrid Columnar Compression (HCC)
• Compressed, columnar format in storage, saving space,
reducing I/O, speeding analytic queries
• Unique: In-Memory Columnar (IMC)
• Extends In-Memory database performance to higher
capacity Flash memory in storage
Fastest Analytics
Storage Index (DRAM)
PMEM Accelerator
*Includes long-running SQL queries, backups, decryption, aggregation, data mining
PCI NVMe Flash
HCC
10:1
Compression
IMC
Exadata
Database Cluster
100g RoCE Network
Exadata
Storage Server
PMEM Accelerator
PCI NVMe Flash
PMEM Accelerator
PCI NVMe Flash
35
SQL
Offload

• Exadata uses frequent heartbeat messages
between nodes to detect possible server failure
• Server failure detection normally requires long
timeout to avoid false server evictions from
cluster
• Hard to quickly distinguish between slow response to
heartbeat due to very high CPU load, and server failure
• Exadata uses RDMA to quickly confirm server
failure
• RDMA uses hardware - remote ports reply even if
software is slow
• 4 RDMA Reads are sent to suspect server, across
all combinations of source and target ports
• If all 4 RDMAs fail, server is evicted from cluster
Exadata RoCE Instant Failure Detection
RDMA
NIC Port #2
NIC Port #2
NIC Port #1
NIC Port #1
36
DB node
failure
TPS with
Exadata
TPS without
Exadata
Sub-second notification vs 1-minute timeout on
non-Exadata platforms

Kernel-based Virtual Machine (KVM): virtualization module in the Linux kernel, allowing the
kernel to function as a hypervisor
• KVM has become the industry standard for Cloud and on-premises deployments
Effective isolation of CPU, memory, OS, and sysadmin for consolidated workloads
• Cross department consolidation, test/dev, non-database applications, deployment in
cloud / hosted environments
Exadata VMs deliver near raw hardware performance
• I/Os go directly to high-speed network bypassing hypervisor
• Combine with unique Exadata network and I/O prioritization to achieve unique full stack
isolation
Trusted Partitions allow licensing by virtual machine
KVM in X9M provides better virtualization metrics compared to OVM (in X8 and earlier)
• Higher VM Memory: up to 2 TB RAM per DB server (max guest VM memory: 1870GB)
• 50% more guest VMs per DB server (12 vs 8) compared to X8
Exadata X9M High Performance Virtualization: KVM
Finance VM
Marketing
VM
Sales VM
For further reference: https://www.oracle.com/a/tech/docs/exadata-kvm-overview.pdf

Applying Machine Learning to Support Real-time Operational Diagnostics
Oracle AHF addresses availability and performance issues in System and
Database Administrator spaces.
• EXAchk specifically developed for Exadata:
• Configuration checks for database, storage and network fabric
switches
- Firmware, Operating System, Exadata Software, Grid Infrastructure,
Database
• MAA Scorecard
- MAA Configuration Review
- Exadata Software Planner
- Exadata Critical Issue alerts
• Automatic Correction (when applicable)
- Configuration Correction
- Critical Issue Avoidance
• Prereq checks for DB/GI upgrades and application continuity
readiness
• Prioritized, continuous improvement based on Development backed
best practices
Oracle Autonomous Health Framework – EXAchk

Blueprint for HA: Designed/Tested Against Failure Scenarios
Exadata Maximum Availability Architecture (MAA)
39
LAN
Within Exadata:
Full Fault Tolerance
Redundant Software
Active clusters,
Disk/flash mirroring
Redundant Hardware
Servers, Disks, Flash,
Network, Power
Redo-based
change
replication
with data
consistency
checking
Local Standby
for HA Failover
Redundant Systems
Redundant Databases
Within a Site:
Local Data Guard
Online patching,
reconfiguration,
expansion
WAN
Across Sites:
Data Guard for DR
Fastest RAC Instance and Node Failure Recovery | Fastest Backup - RMAN Offload to Storage
Deep ASM Integration | Fastest Data Guard Redo Apply | Complete Failure Testing with Shortest Brownouts
Ref. oracle.com/goto/maa
Remote Standby
for Disaster Recovery
Redundant Systems
Redundant Databases

Executive
Summary
Best Database Machine
• Fastest OLTP and Analytics
• High Consolidation Lowers Costs
1
Summary
Exadata X9M
Best Database Cloud Platform
• Consistent across On-premises, Public Cloud and Hybrid
• Enterprise-grade Scalability, Availability and Security
2
Only Platform for Autonomous Database
• Self-driving capabilities lowers costs further
• Ushers in new era of business innovation
3

Konsolidace
Oracle Database
Profesionální služby společnosti
Neit Consulting s.r.o.

42
zadavatel:
projekt: „Dodávka integrovaného databázového systému pro provoz Oracle
databází v ČNB“
předmět: konsolidace Oracle Database na platformě Oracle Exadata
Customer Study

Neit Consulting s.r.o.
+420 720 032 333
Info.cz@neit.group
www.neit.cz
Robert Ernest
Business Development Manager
+420 731 428 393
robert.ernest@neit.group

Now is the time to
consolidate your
databases to the
Oracle Exadata
Let us show you how
44

Oracle databáze – Konsolidovaná Data Management Platforma

Recommandé

Recommandé

Contenu connexe

Tendances

Tendances (20)

Similaire à Oracle databáze – Konsolidovaná Data Management Platforma

Similaire à Oracle databáze – Konsolidovaná Data Management Platforma (20)

Plus de MarketingArrowECS_CZ

Plus de MarketingArrowECS_CZ (20)

Dernier

Dernier (20)

Oracle databáze – Konsolidovaná Data Management Platforma