Tag Archives: Exadata DBA

Exadata Storage Server (Cell Nodes)

Exadata Storage Server (Cell Node)

Exadata Storage Servers, also called Cell Servers, are specialized storage nodes in Oracle Exadata.
Unlike traditional storage, Exadata storage performs database-aware processing such as Smart Scan, Storage Index filtering, and Flash Cache acceleration.

1. What is an Exadata Storage Server?

ComponentDescription
Storage Server (Cell)Physical server storing database data
Cell Server SoftwareSoftware running on storage nodes to manage I/O operations
Smart ScanOffloads SQL processing to storage layer
Flash CacheHigh-speed flash storage used for caching data

2. Exadata Rack Storage Components

ComponentTypical Full Rack Count
Database Servers8
Storage Servers14
Network Switches2

3. Storage Server Processes

ProcessPurpose
CELLSRVMain process handling I/O operations
MS (Management Server)Monitors storage server health
RS (Restart Server)Restarts failed services automatically

4. Smart Scan Feature

Smart Scan allows Exadata to filter rows and columns directly on the storage servers.
Instead of sending full data blocks to the database server, only required rows are returned.

Traditional Storage:
Database Server → Reads full blocks → Filters rows

Exadata Smart Scan:
Database Server → Storage Server filters rows → Sends only required data

5. Storage Index

Storage Index is an in-memory structure maintained by storage servers.
It helps avoid unnecessary disk reads by tracking minimum and maximum column values in storage regions.

6. Flash Cache

Flash UsageDescription
Read CacheCaches frequently accessed data blocks
Write Back CacheImproves write performance
Smart Flash CacheAutomatically manages hot data

7. Exadata Disk Architecture

Physical Disk
      ↓
     LUN 
      ↓
Cell Disk
      ↓
Grid Disk
      ↓
ASM Disk
      ↓
Database Files

8. Useful Storage Server Commands

cellcli -e list cell detail

cellcli -e list celldisk

cellcli -e list griddisk

cellcli -e list flashcache

9. Advantages of Exadata Storage Servers

FeatureBenefit
Smart ScanReduces data transfer and improves query performance
Storage IndexAvoids unnecessary disk reads
Flash CacheHigh-speed data access
RDMA NetworkUltra-low latency communication

Caution: Your use of any information or materials on this website is entirely at your own risk. It is provided for educational purposes only. It has been tested internally, however, we do not guarantee that it will work for you. Ensure that you run it in your test environment before using.

Thank you,
Rajasekhar Amudala
Email: br8dba@gmail.com
Linkedin: https://www.linkedin.com/in/rajasekhar-amudala/

ExaData Network

Oracle Exadata Rack Size

Rack SizeDB ServersStorage ServersTypical Use
Eighth Rack23Dev / Test / DR
Quarter Rack25Small–Medium Production
Half Rack47Medium–Large Production
Full Rack814Large / Mission Critical

Exadata Interconnect Network / Bond Speed (X5 – X11M)

Exadata ModelNetwork TypeSpeed Per PortBonded Speed
X5InfiniBand (FDR)56 Gbps56 Gbps × 2 = 112 Gbps
X6InfiniBand (FDR)56 Gbps56 Gbps × 2 = 112 Gbps
X7InfiniBand (EDR)100 Gbps100 Gbps × 2 = 200 Gbps
X8InfiniBand (EDR)100 Gbps100 Gbps × 2 = 200 Gbps
X8MRoCE (RDMA over Converged Ethernet)100 Gbps100 Gbps × 2 = 200 Gbps
X9MRoCE100 Gbps100 Gbps × 2 = 200 Gbps
X10MRoCE100 Gbps100 Gbps × 2 = 200 Gbps
X11MRoCE100 Gbps100 Gbps × 2 = 200 Gbps

 

Exadata Networking Overview

Oracle Exadata uses a high-speed low-latency internal network connecting
Database Servers, Storage Servers (Cells), and Switches.
This network enables features like Smart Scan, ASM rebalance, RAC cache fusion, and high-speed data transfer.

1. Client Network

FeatureDescription
PurposeApplication connections to the database
TrafficSQL queries, application requests
Network TypeStandard Ethernet (10Gb / 25Gb / 40Gb / 100Gb)
Example FlowApplication → Client Network → SCAN Listener → RAC Node

2. RAC Private Interconnect

The private interconnect network is used for communication between RAC database nodes.

UsageDescription
Cache FusionTransfer blocks between RAC nodes
Cluster HeartbeatMaintains RAC node communication
Parallel QueryData exchange between nodes

 

Exadata ModelNetworking Technology
X2 – X6InfiniBand (40–56 Gb)
X7 – X8InfiniBand (100 Gb)
X8M and LaterRoCE (RDMA over Converged Ethernet)

3. Storage Network (Database → Storage Cells)

Database nodes communicate with Exadata storage servers using a high-speed RDMA network.

FeatureDescription
Smart ScanQuery processing pushed to storage servers
Block ReadsFast data retrieval from storage cells
ASM RebalanceHigh-speed disk rebalance operations
Flash Cache ReadsAccess flash storage with very low latency

Exadata Network Architecture

ComponentTypical Full Rack Count
Database Servers8
Storage Servers12
Network Switches2 (Redundant)

Important Exadata Networking Commands

cat /proc/net/bonding/bondib0

ip addr

rdma link

ibstat

Key Advantages of Exadata Networking

FeatureBenefit
RDMAUltra low latency communication
High BandwidthFaster query processing
Redundant SwitchesHigh availability
Bonded InterfacesNetwork failover and load balancing

Caution: Your use of any information or materials on this website is entirely at your own risk. It is provided for educational purposes only. It has been tested internally, however, we do not guarantee that it will work for you. Ensure that you run it in your test environment before using.

Thank you,
Rajasekhar Amudala
Email: br8dba@gmail.com
Linkedin: https://www.linkedin.com/in/rajasekhar-amudala/

Exadata DB Server (Compute Node) – Memory

Oracle Exadata Rack Size

Rack SizeDB ServersStorage ServersTypical Use
Eighth Rack23Dev / Test / DR
Quarter Rack25Small–Medium Production
Half Rack47Medium–Large Production
Full Rack814Large / Mission Critical

Oracle Exadata DB Server (Compute Node) – Memory (X5 to X11M)

Exadata ModelMemory / DB ServerQuarter Rack (2 DB)Half Rack (4 DB)Full Rack (8 DB)
X5256 GB256 GB × 2 = 512 GB256 GB × 4 = 1 TB256 GB × 8 = 2 TB
X61.5 TB1.5 TB × 2 = 3 TB1.5 TB × 4 = 6 TB1.5 TB × 8 = 12 TB
X71.5 TB1.5 TB × 2 = 3 TB1.5 TB × 4 = 6 TB1.5 TB × 8 = 12 TB
X81.5 TB1.5 TB × 2 = 3 TB1.5 TB × 4 = 6 TB1.5 TB × 8 = 12 TB
X8M1.5 TB1.5 TB × 2 = 3 TB1.5 TB × 4 = 6 TB1.5 TB × 8 = 12 TB
X9M2 TB2 TB × 2 = 4 TB2 TB × 4 = 8 TB2 TB × 8 = 16 TB
X10M3 TB3 TB × 2 = 6 TB3 TB × 4 = 12 TB3 TB × 8 = 24 TB
X11M3 TB3 TB × 2 = 6 TB3 TB × 4 = 12 TB3 TB × 8 = 24 TB

Caution: Your use of any information or materials on this website is entirely at your own risk. It is provided for educational purposes only. It has been tested internally, however, we do not guarantee that it will work for you. Ensure that you run it in your test environment before using.

Thank you,
Rajasekhar Amudala
Email: br8dba@gmail.com
Linkedin: https://www.linkedin.com/in/rajasekhar-amudala/

Exadata DB Server (Compute Node)- CPU

Oracle Exadata Rack Size

Rack SizeDB ServersStorage ServersTypical Use
Eighth Rack23Dev / Test / DR
Quarter Rack25Small–Medium Production
Half Rack47Medium–Large Production
Full Rack814Large / Mission Critical

 

Oracle Exadata DB Server (Compute Node) – CPU (X5 to X11M)

Exadata ModelCPU (Per DB Server)Total (Per DB Server)Total CPU (Quarter Rack – 2 DB Servers)Total CPU (Half Rack – 4 DB Servers)Total CPU (Full Rack – 8 DB Servers)
X52 × 18-core Intel36 cores36 × 2 = 72 cores36 × 4 = 144 cores36 × 8 = 288 cores
X62 × 22-core Intel44 cores44 × 2 = 88 cores44 × 4 = 176 cores44 × 8 = 352 cores
X72 × 24-core Intel48 cores48 × 2 = 96 cores48 × 4 = 192 cores48 × 8 = 384 cores
X82 × 24-core Intel48 cores48 × 2 = 96 cores48 × 4 = 192 cores48 × 8 = 384 cores
X8M2 × 24-core Intel48 cores48 × 2 = 96 cores48 × 4 = 192 cores48 × 8 = 384 cores
X9M2 × 32-core Intel64 cores64 × 2 = 128 cores64 × 4 = 256 cores64 × 8 = 512 cores
X10M2 × 96-core AMD192 cores192 × 2 = 384 cores192 × 4 = 768 cores192 × 8 = 1536 cores
X11M2 × 96-core AMD192 cores192 × 2 = 384 cores192 × 4 = 768 cores192 × 8 = 1536 cores

Caution: Your use of any information or materials on this website is entirely at your own risk. It is provided for educational purposes only. It has been tested internally, however, we do not guarantee that it will work for you. Ensure that you run it in your test environment before using.

Thank you,
Rajasekhar Amudala
Email: br8dba@gmail.com
Linkedin: https://www.linkedin.com/in/rajasekhar-amudala/

Exadata Part Replacement Runbook

Exadata Part Replacement – Quick Notes

Hot Swappable Components

ComponentReplacement Type
LUN / DiskHot Swappable
FanHot Swappable
Single Power SupplyHot Swappable
Flash Disk (From X7)Hot Swappable
DIMMShutdown Required
MotherboardShutdown Required

Before Replacement (If Shutdown Required)

**** Always verify ASM rebalance activity before making grid disks inactive.
select * from gv$asm_operation;
cellcli -e list griddisk attributes name, asmmodestatus, asmdeactivationoutcome, status 
cellcli -e alter griddisk all inactive 
cellcli -e list griddisk attributes name, asmmodestatus, asmdeactivationoutcome, status 
service celld status 
service celld stop 
shutdown -h now

After Replacement

service celld status 
cellcli -e list griddisk attributes name, asmmodestatus, asmdeactivationoutcome 
cellcli -e alter griddisk all active
cellcli -e list griddisk attributes name, asmmodestatus, asmdeactivationoutcome

select * from gv$asm_operation;

Caution: Your use of any information or materials on this website is entirely at your own risk. It is provided for educational purposes only. It has been tested internally, however, we do not guarantee that it will work for you. Ensure that you run it in your test environment before using.

Thank you,
Rajasekhar Amudala
Email: br8dba@gmail.com
Linkedin: https://www.linkedin.com/in/rajasekhar-amudala/