Showing posts with label Flash Cache. Show all posts
Showing posts with label Flash Cache. Show all posts

Remove Exadata Flash Grid disk and configure as normal Flash

In previous blog we have created Grid Disk on Flash Disk. Here we are removing Flash Grid Disk and configuring normal Flash Cache.

Step-1

Remove or move database or any contents to some another ASM Disk group

Step-2

Drop ASM disk group which was create on flash disk

Step-3

Drop all the flash grid disk

CellCLI> drop griddisk all prefix=flashdisk force

GridDisk FlashDisk_FD_00_exa1cel01 successfully dropped
GridDisk FlashDisk_FD_01_exa1cel01 successfully dropped
GridDisk FlashDisk_FD_02_exa1cel01 successfully dropped
GridDisk FlashDisk_FD_03_exa1cel01 successfully dropped
GridDisk FlashDisk_FD_04_exa1cel01 successfully dropped
GridDisk FlashDisk_FD_05_exa1cel01 successfully dropped
GridDisk FlashDisk_FD_06_exa1cel01 successfully dropped
GridDisk FlashDisk_FD_07_exa1cel01 successfully dropped

Step-4

Configure Exadata Flash Disk as Grid Disk

We can use Flash Cache as Grid Disk to create the ASM disk group and can keep the database on flash disk but this is not reccomended solution as flash cache performs better to cache frequently access data rather than creating database on it.

If you want put your database on Flash cache than below are the steps to configure FLASH CACHE as ASM disk.

Steps:

1. Drop existing flash cache
2. Create flashcache with required size
3. Create grid disk from Flash Cache
4. Create ASM disk group on top of the Flash Griddisk
5. Put your database on ASM disk group which has been configured from Flash grid disk

Step-1

CellCLI>drop flashcache

If it gives error like below

CELL-02769: Cannot complete flash cache drop on: FD_03_ex01cceladm03. 
Received error: CELL-02732: There may be data not synchronized to the grid disks (dirty data) in flash cache.
Use ALTER FLASHCACHE to flush the dirty data and to stop caching, and then retry the command.

This error comes because flash cache mode has been configured as writeback hence some of the dirty data are still not written to the grid disks.

To overcome from this error flush all the flashcache

CellCLI>  alter flashcache all flush
Flash cache exa1cel01_FLASHCACHE altered successfully

CellCLI> drop flashcache 
Flash cache exa1cel01_FLASHCACHE successfully dropped 

Step-2

Exadata Smart Flash Caching

Smart Flash Cache:
  • Primary task of smart flash cache is to hold frequently accessed data in flash cache so next time if same data required than physical read can be avoided by reading the data from flash cache.
  • Exadata caching mechanism focuses on caching frequently accessed data and index block with performance critical information like control file and file headers. 
  • Also we can influence caching priorities by using CELL_FLASH_CACHE storage attribute for specific database objects
What will be cached ?
  • Frequently accessed data
  • Index Block
  • File header reads and writes
  • Control file reads and writes
  • DBA can set the caching priority
What will be not cached?
  • Backup related I/O
  • Database pump I/O
  • Data file formatting
  • I/O to mirror copies
  • Table scan don't control the cache

Cellcli - Help

CellCLI> help
HELP [topic]
    Available Topics:
        ALTER
        ALTER ALERTHISTORY
        ALTER CELL
        ALTER CELLDISK
        ALTER GRIDDISK
        ALTER IBPORT
        ALTER IORMPLAN
        ALTER LUN
        ALTER PHYSICALDISK
        ALTER QUARANTINE
        ALTER THRESHOLD
        ASSIGN KEY
        CALIBRATE
        CREATE
        CREATE CELL
        CREATE CELLDISK
        CREATE FLASHCACHE
        CREATE GRIDDISK
        CREATE KEY
        CREATE QUARANTINE
        CREATE THRESHOLD
        DESCRIBE
        DROP
        DROP ALERTHISTORY
        DROP CELL
        DROP CELLDISK
        DROP FLASHCACHE
        DROP GRIDDISK
        DROP QUARANTINE
        DROP THRESHOLD
        EXPORT CELLDISK
        IMPORT CELLDISK
        LIST
        LIST ACTIVEREQUEST
        LIST ALERTDEFINITION
        LIST ALERTHISTORY
        LIST CELL
        LIST CELLDISK
        LIST FLASHCACHE
        LIST FLASHCACHECONTENT
        LIST GRIDDISK
        LIST IBPORT
        LIST IORMPLAN
        LIST KEY
        LIST LUN
        LIST METRICCURRENT
        LIST METRICDEFINITION
        LIST METRICHISTORY
        LIST PHYSICALDISK
        LIST QUARANTINE
        LIST THRESHOLD
        SET
        SPOOL
        START

Sample Questions for Exadata Certification 1

Exadata Implementation Specialist Examination (1Z0-536)

1. Which types of data are most likely to be cached In the Exadata Smart Flash Cache?

A. Results of random reads.
B. Results of table scans.
C. Write to a mirror
D. Redo data
E. All data is cached In the Flash Cache

2. Your customer wants to use Hybrid Columnar Compression to get maximum compression on their data. Which option represents a best practice for achieving this goal?

A. Load compressed files with Direct Path loading
B. Sort incoming data on a column with a low cardinality
C. Use DBFS and external flies to load the data
D. Use Data Pump to load data

3. Which statement would you make about sharing I/O resources in an Exadata environment?

A. You can manage workloads within a database with Database Resource Manager.
B. You can manage workloads across multiple databases with Database Resource Manager.
C. You can manage workloads within a database with 10 Resource Manager.
D. You cannot manage workloads across multiple databases.

4. How do you back up the software used on an Exadata storage cell?

A. You use RMAN.
B. You back up the software files using an operating system method.
C. You do a complete copy of their system storage area.
D. You do not need to backup this software, as there is an automated recovery mechanism built into each cell.

5. Which statement properly describes potential allocation for a grid disk?

A. A grid disk must use all of a cell disk.
B. A grid disk can only use portions of a cell disk across all cells.
C. A grid disk can use portions of a cell disk on a subset of cells.
D. A grid disk can span multiple Exadata Storage Server ceils.

History of Exadata

First Oracle's exadata was released in 2008 known as hardware and software engineered to work together. Since than we have seen 6 release of exadata machine with improved the combination of hardware and software. Latest X5-2 released in 2015. Here we have given short history of each version from V1 to X5-2.



V1: The first Exadata was released in 2008 and labeled as V1. It was a combination of HP hardware and Oracle software. There were no flash cache used in this version. At that time Exadata was made specially for Data ware housing only. This product was not accepted widely as it was suffered from overheating. 


V2: After V1 Oracle has launched Exadata V2 in 2009. It was a combination of Sun hardware and Oracle software. Here in V2 Oracle has upgraded storage cell with 384 GB of Exadata smart cache. Many o the customer has changed the Exadata from V to V2 because of the overheating problem in Exadata V1.

X2: This is the third edition of Oracle Exadata released in 2010.There are two version X2-2 stands for dual core CPUs and X2-8 stands for 8 core CPUs database server. This version is supporting both OLTP and Data warehouse. Also this has massive features compare to older Exadata in terms of CPUs, Memory, Compression, Smart scan etc...


X3: This is the forth edition of Exadata released in 2012. There is no much more difference between X2 and X3 except in CPU and RAM. This version has more RAM and CPU power then the previous one.

X4: It was released in 2013. In this version processing was increased to 2×12 core CPUs. It had the capacity to upgrade memory to 512GB in a compute node. Flash and disk storage was also increased. It featured a new model of high-capacity disk. 600GB disks were retired, and included 1.2TB, 10,000 RPM disks. These disks were a smaller form factor (2.5” vs 3.5”). The other big change with the X4-2 was the introduction of an active/active InfiniBand network connection. On the X4-2, Oracle broke the bonded connection and utilized each InfiniBand port independently. This allowed an increased throughput across the InfiniBand fabric.

X5: It was announced in early 2015, the sixth generation of Exadata. The X5-2 was a dramatic change in the platform, removing the high-performance Disk option in favor of an all-flash, NVMe (Non-Volatile Memory Express) model. High-capacity disk sizes remain the same at 4TB per disk. The size of the flash cards doubled to 6.4TB per storage server. Memory stayed consistent with a base of 256GB, upgradeable to 768 GB, and the CPU core count increased to 18 cores per socket. With X5, customers are allowed to purchase X5-2 with any configuration required—a base rack begins with two compute nodes and three storage servers.

Exadata Features



Let's make a note of major features of exadata. We will see all the features in depth in next couple of posts.


1. Smart Scan / Offloading
2. Storage Index
3. Hybrid Columnar Compression (HCC)
4. Smart Flash Cache
5. IORM