9. Leverage Azure with IaaS options or migrate to a fully managed database for
more benefits
IaaS
Azure Virtual
Machines (IaaS)
Enterprise ready Lower TCO Built-in security
Integrated with Azure data ecosystem
Proven resilience & stability
Always on Azure
Fully managed databases
Azure Database
for PostgreSQL
Azure SQL Azure
Cosmos DB
Get more with
fully managed databases
Fully
managed
Intelligent
Performance
Highly
scalable
High
Availability
Maximum
control
10. Oracle applications supported on Azure
• Microsoft Azure offering
called FlashGrid Node for
a third-party supported
Oracle RAC solution
• OS Clustering solutions
to create similar
performing options to
Oracle RAC One-Node in
a public cloud.
BYOL license for Oracle
Database on Azure
Infrastructure-as-a-Service
• JD Edwards EnterpriseOne
versions 9.2 and above
• E-Business Suite
• WebLogic Server
• Peoplesoft
• Essbase
• Hyperion
Enterprise applications supported
running on Oracle Linux or other
supported operating systems
High availability and
disaster recovery options
Customers can choose to bring
other Oracle Databases images
as well.
DATA GUARD
22. When an
Exadata
Migrates
• Purchase of Exadata was most likely
made due to 1-2 large databases.
• Rest of databases were consolidated to
make the most of the investment.
• Due to the Exadata features, there will
be hesitancy and one database will be
chosen for a POC to assess and migrate
to a public cloud.
• One Database assessment grants a view
into much of the Exadata workloads.
23. What is Oracle Exadata?
• An Engineered hardware and software solution specifically designed to run Oracle workloads
• Consists of database and cell nodes, (servers), infiniband, (fast network), PDU, (power) in a rack configuration.
• Proprietary features:
• Offloading, (using cell nodes to perform high IO workloads instead of the primary database nodes.
• Storage indexes, (indexes in memory only.)
• Result cache for faster buffer results
• Flash cache and flash logging.
• HCC, (Hybrid Columnar Compression) for faster scans of compressed data
• Can use Oracle RAC or instance caging for single database
• Software intelligence installed to know when to use Exadata features to benefit the workload
24. Can Exadata be Built in
Azure?
As Exadata is an engineered system, this solution can’t be run
inside an Azure datacenter, but-
• With Oracle Dataguard as a single Oracle instance with the
correctly scaled VM
• With Oracle Dataguard instead of RAC or use Flashgrid to use
a vendor supported solution.
• Using Goldengate or another migration with a peer-to-peer
sync solution, dual databases.
• Load balancing between Oracle database
instances.
• materialized views
• Index additions
• Compression storage solutions.
25. How is Exadata Different?
Exadata Feature Oracle from Exadata to Azure
The Oracle Exadata is an engineered system for Oracle workloads that delivers the
highest levels of database performance available. An Exadata is a “cloud in a box”
composed of database servers, Oracle Exadata Storage Servers, an InfiniBand fabric
for storage networking and all the other components required to host an Oracle
Database, along with engineered features such as cell offloading for increased IO
performance, storage indexes, Hybrid Columnar Compression, (HCC), flash cache
and logging, storage snapshots, thin cloning and other features that will benefit
workloads designed to use them.
This will be replaced with Azure VM and customers can choose what is best
suitable for their workload. Identifying the workload demand is essential and it’s
crucial not to attempt to lift and shift the Exadata hardware- it’s not a lift and
shift option, but the database workloads must be “decoupled” from the
engineered system.
Patching – Managed by Oracle contract
Patching – As it is an IaaS based solution, customers will be responsible for Oracle
level Patching, but can use Oracle Enterprise Manager lifecycle management pack
to automate OS and Oracle patching.
Cell Offloading - The storage in the Exadata Database Machine is not just dumb
storage. The storage cells are intelligent enough to process some workload inside them,
saving the database nodes from that work. This process is referred to as cell offloading.
Storage options in Azure such as Azure NetApp Files , Silk or Elasticsan can
supplement the IO demands that Exadata workloads require.
Storage indexes that will need to be physically created that currently only exist in
memory for smart scans. Smart Scan is a subset of Cell Offload Processing and is
perhaps the feature that is most highly associated with Exadata. Smart Scan allows
Oracle databases to push query information down to the storage level, specifically:
Filtration (predicate information)
Storage Indexes aren’t possible in Azure, but physical Indexes and automatic
indexing in 19c can assist in creating what may be missing from the workload.
Flash cache can run in Azure, Ultra disk can be used for flash logging solutions.
Hybrid Columnar Compression (HCC) enables very high level and high-
performance data compression for EDW Workloads. HCC organizes data into sets
of rows called compression units. Within a compression unit, HCC organizes the
data by column and then compresses the data.
No HCC support since it is an Exadata specific feature – Needs to be replaced
with other compression options (i.e. Advanced Row Compression) Silk has
significant compression and dedupe options that can rival HCC for Oracle
workloads in Azure and ANF offers close compression, too.
Oracle Exadata rarely houses ONE database, but most often has numerous
databases with a minority using the majority of Exadata benefiting features.
Identify what databases are not using the Exadata features for a lift and shift
option. The few that are should be decoupled carefully from the engineered
system.
26. WANT TO LEARN MORE?
• ORACLE ON AZURE IAAS WHITE PAPER
• MICROSOFT DATA ARCHITECTURE BLOG
• ORACLE ON AZURE GITHUB
• MICROSOFT LEARN- ORACLE ON AZURE