Oracle Technical Q&A Guide
Mastering Exadata, Data Guard, and Oracle Cloud Infrastructure
⚡ Oracle Exadata Highlights
Exadata is a pre-configured combination of hardware and software which provides a platform to run the Oracle Database.
- DB Server
- Cell Storage
- Infiniband Switch
- Cisco Switch
- PDU
- Smart Scan
- Smart Flash Cache
- IORM
- Storage Index
- EHCC (Exadata Hybrid Columnar Compression)
Exadata comes in the following configurations: Full Rack, Half Rack, Quarter Rack, and 1/8th Rack.
Storage Indexes consist of a minimum and a maximum value for up to eight columns. This structure is maintained for 1MB chunks of storage (storage regions). Storage Indexes are stored in memory only, are never written to disk, and filter out data from the consideration.
SNMP
No
KEEP, DEFAULT, and NONE
512MB per module. Each storage cell has 4 modules, making it 4X512 MB per CELL.
The flash cache is a hardware component configured in the Exadata storage cell server which delivers high performance in reading and writes operations. The primary task is to hold frequently accessed data in the flash cache so physical reads can be avoided on subsequent data requirements.
cellcli is a cell command-line utility used to manage cell storage. It is a utility to administer cell storage, providing commands like CELLCLI-ALTER, CELLCLI-CREATE, CELLCLI-DROP, and CELLCLI-LIST.
- Query Low
- Query High
- Archive High
- Archive Low
DBMS_COMPRESSION
- MS: Management Server
- cellsrv: Cell Server
- RS: Restart Server
All HDDs are hot-swappable, so if proper redundancy is used, the HDD can be directly removed and replaced. Storage software will handle everything else in the background.
12 disks
The spine switch is used to connect or add more Exadata machines to the cluster.
Multiple methods are available depending on source setup and downtime limits:
- Export/Import
- Physical Standby
- Logical Standby
- Transportable Tablespace / Transportable Database
- Golden Gate
- RMAN cold and hot backup restoration
- Oracle Streams
- Stop Database and Listener
- Stop Cluster
- Shutdown Database Servers
- Shutdown Cell Storage
- Shutdown all the switches
- Remove power from PDUs
Start-up Procedure: It is executed in reverse order.
2
14
Yes
iDB protocol
Database servers support either Linux or Solaris (decided at configuration), while Cell storage only runs Linux.
ASR stands for Auto Service Request. It is a tool used to manage Oracle hardware, automatically raising an SR with Oracle Support and notifying the customer when a hardware fault occurs.
- Install OEM agent on DB server
- Launch auto-discovery with the use of One Command XML file
- Specify the required credentials for all the components
- Review Configuration
- Complete the setup
It can be done through the ILOM of the DB or Cell server.
The CELLIP.ORA file contains the list of storage servers accessed by the DB server.
- Exacheck
- sundiagtest
- oswatcher
- OEM 12c
EHCC stands for Exadata Hybrid Columnar Compression, which is used to compress data in the Database.
It refers to SQL processing being moved from the database layer to the storage layer. The primary benefit is that it significantly reduces the volume of data returned to the database server, bypassing major database bottlenecks.
Cellcli is only used on a respective cell storage server. dcli (Distributed Command Line Utility) is used to execute commands across multiple storage or DB servers simultaneously.
IORM stands for I/O Resource Manager. It manages and prioritizes the I/Os of multiple databases on a storage cell.
Run Exacheck and evaluate the best practice setup.
- Public/Client Network: For Application Connectivity
- Management Network: For Exadata H/W management
- Private Network: For cluster interconnectivity and storage connectivity
SQL> alter table table_name move compress for query high;
Manual backups are not required; Exadata automatically uses an internal USB drive (Cell boot Flash Drive) to back up storage software.
Write-through: Flash cache is used only for reading operations.
Write-back: Flash cache is used for both reading and writing operations.
Flash Cache
40 Gbps
High Capacity: More storage space but lower RPM (7.5k).
High Performance: Less storage space but higher RPM (15k).
Before and after making any configuration change on the Database Machine.
Grid Disks are created on top of Cell Disks and presented to Oracle ASM as ASM disks. Space is allocated in chunks from outer to inner tracks of the Cell disk. Multiple Grid Disks can exist per Cell disk.
Infiniband Network
- Create Directory
- Create Tablespace on a database intended for DBFS
- Create a user for DBFS
- Grant required privileges to the created user
- Connect to the database using that user
- Create the DBFS file system using
dbfs_create_filesystem_advanced - Mount the file system by starting
dbfs_client
- Golden Gate
- Transportable Tablespace
- Incremental Transportable Tablespace
- Data Pump
Cell and Grid Disks are logical components of physical Exadata storage. A Storage Server Cell consists of disk drives put together to store data. Each Cell Disk corresponds to a LUN formatted by the storage software, typically housing 12 disk drives.
It consists of four 96GB PCIe flash cards on each Storage Server cell providing 384GB total per cell. It provides high-speed access to data and reduces latency by fetching from flash memory rather than mechanical disks.
A software feature that enhances database performance. It processes queries intelligently by filtering data directly at the storage layer based on criteria and only returning specific rows instead of entire blocks. It also utilizes column projection to return only requested columns.
The PARALLEL_FORCE_LOCAL parameter can be set at session level for jobs.
Traditional Export/Import, Tablespace transportation, or GoldenGate Replication after data restore.
Exadata runs Oracle Linux OS and has Solaris available on its engineered systems.
Cell flash cache read hits
+DBFS_DG
There are 53 Exadata-specific wait events.
The X3-8 consists of 2 large SMP compute servers while the X3-2 scales up to 8 compute servers to match scaling needs.
Database Instance, ASM Instance, and Database Resource Manager.
DBRM is a database instance feature, while IORM is a storage server software feature.
Exadata configurations come as Full Rack, Half Rack, or Quarter Rack:
- Full Rack (X2-2): 8 Database Server nodes (Intel Xeon 5670 processors with 6 CPUs and 96GB memory per node) and 14 storage cells.
- Half Rack: Half the capacity with 4 Database Server nodes and 7 storage cells.
Software Level: Smart Scan, Flash Cache, Hybrid Columnar Compression, and IORM.
Approximately 3 hours per Cell and DB server. Infiniband & PDU patching takes roughly 1 hour each.
It minimizes the risk of storage disk failures.
Rolling:
- Advantages: No downtime required.
- Disadvantages: Takes more time. Disk failures on a separate cell under normal redundancy can lead to corruption.
Non-Rolling:
- Advantages: Requires less time. Eliminates disk failure risks under normal redundancy.
- Disadvantages: Complete downtime required.
- AUTO_MANAGE_MAX_ONLINE_TRIES: Controls the maximum attempts to bring a disk online.
- AUTO_MANAGE_EXADATA_DISKS: Directly enables/disables auto disk management.
- AUTO_MANAGE_NUM_TRIES: Controls the attempts for performing automated tasks.
CellCLI> ALTER CELL flashCacheCompress=true
14 storage nodes
🛡️ Oracle Data Guard Highlights
It keeps data protected and available, offloads backup work to standby instances, detects gaps in logs automatically, and streamlines role transitions using the Guard Broker.
Redo Transport Services, Apply Services (Log Apply), and Role Transition Services.
Maximum Availability, Maximum Protection, and Maximum Performance.
Maximum Performance.
It doesn’t affect primary database performance while providing the highest potential data protection. Transactions commit once written to the online redo logs.
It completely guarantees zero data loss. If the primary database is unable to write redo to a standby instance, it will shut down to prevent processing un-mirrored data.
Redo must be written to both the online log and the standby database before a transaction commits. This protects data even if the primary server fails.
It optimizes both availability and safety. Redo must be confirmed on both the primary and standby databases before committing; however, if the standby site goes down, the primary stays running.
Use the alter database command, assign the standby database, and select from the available performance modes.
Depending on your specific Oracle version, it ranges from 9 to 30.
- Enable forced logging
- Create a password file
- Configure standby redo logs
- Ensure archiving is enabled
- Initialize primary DB parameters
- Configure networking/nodes for proper communications
Provides higher availability, handles offloaded backups and reporting, keeps the standby data secure, and ensures fast disaster recovery.
- Log Writer: Collects info and writes it to online redo logs locally and standby sites remotely.
- Archive Process: Copies online redo logs locally or to remote standbys.
- Fetch Archive Log (FAL): Serves archive log requests from FAL client processes on standby sites.
- Log Network: Establishes network links to the standby servers.
- Fetch Archive Log (FAL): Pulls missing sequences from primary when gaps occur.
- Remote File Server (RFS): Receives incoming redo files from primary.
- Archive Process: Archives redo files applied on standby.
- Managed Recovery (MRP): Applies incoming redo log data to standby databases.
Modify the initialization parameters on the primary site to explicitly specify a delay in minutes (e.g., 60' for one hour). Redo data is still transferred immediately but won’t be applied until the timer expires.
It manages the automated transmission of redo logs to the standby sites, detects sequence gaps when network disconnects happen, and retrieves replacement logs to resolve gap issues automatically.
Physical databases apply changes at the block level using Redo Apply, while logical databases use SQL Apply to convert incoming redo data into SQL statements that execute directly against the standby database.
The Broker is a management framework offering command-line and graphical interfaces integrated with Enterprise Manager. It simplifies failover/switchover orchestration.
Create and enable configurations, administer configuration across different hosts, monitor operational health, and invoke automated fast-start failovers without needing human operators.
A database synchronized using SQL statements. The database structure can be modified at later points, and tables being synced from the primary remain read-only, while custom indices and tables can be created for reporting and optimization purposes.
Certain data types and tables cannot be synchronized easily with SQL Apply.
Yes, but automated Data Guard features are omitted. You can ship logs and write custom apply scripts manually.
It’s highly advantageous for enterprise firms handling massive volumes of business data. Certifications showcase skills in data recovery, avoiding database corruption, and disaster risk mitigation.
Services that update the standby instance with the redo log records coming from primary instances to maintain synchrony. Data can be read from archive files or standby files immediately when real-time apply is configured.
☁️ Oracle Cloud Infrastructure (OCI)
Oracle Cloud Infrastructure (OCI) offers highly available, modular compute and storage services on top of a virtualized secure network.
OCI provides a cloud platform to build, migrate, and run various applications in a high-performance environment.
- Virtual Cloud Network (VCN)
- Subnet
- Virtual Network Interface Card (VNIC)
- Security Lists
- Route Tables
- Route Rules
A virtual network appliance in a regional subnet used to securely link back-end services (partner APIs, internal functions) to callers.
Standard package typically describes tariffs including core capabilities like free data allowances, basic services, rentals, deposits, and call/data transfer fees.
No, Oracle includes Transparent Data Encryption in all base, regular, enterprise, high-performance, and extreme packages.
- Form module
- Menu module
- PL/SQL Library module
- Object Library module
Used to identify migrated and chained rows, validate object structures, collect usage statistics for the query optimizer, and delete object statistics from the dictionary.
It assists with database recovery and maps exactly where the database and redo log files are located so they can be opened on startup.
A varying length binary string data type that stores up to 2 gigabytes of raw data in byte lengths.
A backup that reads and writes actual database records into files via Export utilities and restores them using Import tools.
Occurs when processes repeatedly query a data dictionary table due to limited capacity inside the data dictionary cache.
.RDF and .REP
A storage optimization technique where a hash function is applied to row cluster keys for storage in hash tables to speed up data retrieval.
Security/User assessments, Data Discovery, Data Masking, Auditing, Backup & Recovery, Scalability, and High Availability.
The customer takes responsibility for software installation/maintenance inside virtual images, complete DBA/root level access, automated database backups, and performing recovery tasks when failures happen.
The logical storage unit of a database that groups together database structures. Data is physically stored in data files associated with the tablespaces.
Formed automatically upon any database creation, it holds all of the data dictionary tables required by Oracle and must stay active constantly.
Also called materialized views, snapshots duplicate master data to non-master sites in synchronized environments or cache expensive analytical query results.
A way to break down large transactions into tiny parts so that transactions can be partially rolled back to a specific SAVEPOINT if errors occur.
Library Cache: Stores information regarding SQL statement parsing, cursors, and execution plans.
Data Dictionary Cache: Stores user privileges, segment info, and account information.
Tables (arrangements of rows and columns), Tablespaces (logical storage units), Views (virtual tables), and Indexes (tuning methods for record searches).
Post-database triggers fire after Oracle Forms issues commits, while post-form triggers execute during the posting phase after a database commit finishes.
Navigate to the OCI console, click Identity & Security, select Compartments, click on Actions next to the target compartment, and select Delete Compartment.
A DRG behaves as a virtual cloud router, linking your on-premises network securely to VCNs or routing data across multiple VCN attachments.
A tool package that transforms sensitive production data into safe formats and creates smaller subsets of that data for use in testing and partner environments.
Analyzes database traffic to block incoming threats and aggregates security audit logs across operating systems, directories, and databases.
Used by owners to directly configure and manage security credentials in their respective Oracle wallets.
Hot Backup: Conducted while the database remains open and accessible.
Cold Backup: Conducted only after the database has shut down using standard commands.
A managed asset deployed and automated using the OCI Resource Manager.
Using a unique identifier string called an Oracle Cloud Identifier (OCID).
Oracle Cloud’s next-generation security suite that keeps data isolated and audits all actions to customer databases.
Yes; compartments can be nested up to six levels deep.
In Oracle Cloud Object Storage, which provides thousands of terabytes of scalable cloud space for backups.
Oracle database cloud service operates like a standard single-instance database, but Oracle supplies the computational resources, hosting, and power.
Oracle provides complete Cloud Database as a Service (DBaaS) solutions optimized for enterprise database testing, staging, and development workloads with high availability and built-in scaling.
Leave a comment