Lore Data - Tecnologia em Banco de Dados

Replacing a physical disk in the Exadata storage server

Wed, 26 Apr 2023 06:25:57 -0700

I'll quickly show here the procedure for replacing a Physical Disk in predictive failure status in the Storage Cell.

Check which disk is failed:

List the LUN to see more details and make sure that is not used for Cell OS:

Check if there is any griddisk that cannot be deactivated. If any griddisk is listed then you cannot proceed before fixing the issues that are preventing the griddisks to become inactive.

List the disk again before the replacement:

Drop the physical disk for replacement and then inform the Field Engineer onsite that he is ok to proceed with replacing the disk in the storage server:

Oracle Field Engineer removes the physical disk from the disk slot:

The Field Engineer inserts the new disk into the slot and you are not able to see the disk by its previous serial number anymore:

You can query using the enclosure and slot numbers which makes the name of the disk:

Happy disk replacement.

Franky

Replacing a physical disk in the Exadata compute node

Tue, 25 Apr 2023 18:24:31 -0700

Quick blog post showing how to replace a physical disk in an Exadata compute node.

Check the disk enclosure:

[root@cmx1db02 ~]# /opt/MegaRAID/MegaCli/MegaCli64 -encinfo -a0
Number of enclosures on adapter 0 -- 1

Enclosure 0:
Device ID : 252
Number of Slots : 8
Number of Power Supplies : 0
Number of Fans : 0
Number of Temperature Sensors : 0
Number of Alarms : 0
Number of SIM Modules : 1
Number of Physical Drives : 4
Status : Normal
Position : 0
Connector Name : Unavailable
Enclosure type : SGPIO
FRU Part Number : N/A
Enclosure Serial Number : N/A
ESM Serial Number : N/A
Enclosure Zoning Mode : N/A
Partner Device Id : Unavailable

Inquiry data :
Vendor Identification : LSI
Product Identification : SGPIO
Product Revision Level : N/A
Vendor Specific...Read More

Exadata Disk architecture demystified

Tue, 06 Dec 2022 06:13:35 -0800

When working with Oracle databases running on Exadata we have to understand the storage and disk architecture to be able to better deliver our work.

An Exadata Database Machine is a rack made of Database Servers, Storage Servers, Power Distribution Units and Switches.

The database and ASM instances run their processes and allocate their memory areas in the Database Servers. On the other hand we have the Storage Servers which provide the disks which combined mount the ASM diskgroups available to the databases to store data. The communication between Database Server and Storage Servers is made through Infiniband Switches until the X8 generation and starting with X8M that communication is made through RoCE (RDMA (Remote Direct Memory Access) over Converged Ethernet) Switches.

In this blog post we are going to explore the disk architecture and the relationship between the ASM Diskgroups and the Storage Server disks.

Storage Servers, depending on the generation of the hardware and/or customer choice, considering the latest generation (X9M) these are available:

For this blog post let's consider the Storage Server High Capacity since it is the one vastly used. It has twelve hard disk drives, four NVMe flash cards and twelve persistent memory cards.

Each of those we call PhysicalDisk which will then have a CellDisk created on it. CellDisks will have partitions to make GridDisks, FlashCache, FlashLog, PersistentMemoryCache, PersistentMemoryLog.

GridDisks are created by default on the CellDisks that are from the HDDs, FlashCache and FlashLog are created on the CellDisks that are from the flash cards and lastly PersistentMemoryCache and PersistentMemoryLog are created on top of the CellDisks that are from the pmem cards.

Running the command Migrated 6 11.2.0.4 DBs from Exadata X4 to Exadata X8 Cloud at Customer Gen2 by upgrading and converting to multitenant in 19c with a low downtime window.

The strategy adopted was a combination of physical and logical migration using incremental backups and transportable tablespaces to keep the downtime to be less than 3 hours.
The client didn’t have enough storage space to provide a staging area for the migration so I used part of the new Exadata storage by creating an ACFS filesystem and exported it via HANFS to use it as the unified staging area for the migration. Had to tune the mounted NFS filesystem to achieve the best performance possible.

Deployed three Exadata X9M appliances (Prod, DR and Dev/QA) before it was officially released and announced by Oracle for a power and gas utility company in the US.

Implemented a disaster recovery solution with two ZDLRA X8M + one StorageTek SL8500 to protect 6 very large databases for a government agency in Brazil.

The strategy adopted was to backup the primary databases to a ZDLRA in DC-A and then have its backups copied to the SL8500 while maintaining the standby databases in sync with their primaries and back them up to another ZDLRA in...Read More

How to enable RAM Cache in the Exadata Storage Servers

Sat, 03 Oct 2020 06:32:31 -0700

A brief background on In-Memory OLTP Acceleration or RAM Cache

In this brief blog post a new feature introduced in the Exadata Storage Server image 18.1 will be presented. You probably heard a lot about the new Exadata X8M and its new Intel Optane DC Persistent Memory (PMem) in the storage server. This new feature+architecture allows the database processes running on the database servers to remotely read and write through a protocol called Remote Direct Memory Access (RDMA) from/to the PMem cards in the storage servers. RDMA exists in the Exadata architecture since the beginning, so this is, by far, not new:

What you probably do not know is that there is a feature called In-Memory OLTP Acceleration (or simply RAM Cache) which was introduced in the Exadata Storage Server image 18.1.0.0.0 when the Exadata X7 was released. That feature allows read access to the Storage Server RAM on any Exadata system (X6 or higher) running that version or above. It is not the same thing as PMem since RAM is not persistent, but it is indeed very cool since you could take advantage of the RAM available in the Storage Servers.

Modern generations of Exadata Storage Servers come with a lot of RAM available. X8 and X7 come with 192GB of RAM by default while X6 used to come with 128GB of RAM.

Unfortunately this feature is only available on Storage Servers >=X6 and these are the requirements:

Oracle Exadata System Software 18c (18.1.0)
Oracle Exadata Storage Server X6, X7 or X8
Oracle Database version 12.2.0.1 April 2018 DBRU or 18.1 or higher

That large amount of RAM is rarely fully utilized by the Exadata Storage Servers. Therefore with this RAMCache feature you can use all or part of the available RAM in the Storage Servers to extend your database buffer cache to the Storage Servers RAM for reading operations.

In the new Exadata X8M the I/O latency is under 19µs for read operations due to the PMem cache combined...Read More

Live streaming: Oracle vs SQL Server

Thu, 23 Apr 2020 17:47:34 -0700

Olá pessoal. Fui convidado de última hora para participar da Live que vai acontecer amanhã, 24 de Abril de 2020 às 20:00, onde conversaremos pacificamente sobre Oracle e SQL Server. A ideia é mostrar o que os 2 SGBDs mais usados no mundo corporativo tem de melhor para oferecer e também alguns dos seus pontos fracos.

A discussão promete ser quente e de alto nível, pois temos no time Oracle o Ricardo Portilho Proni, o Everton Monteiro e eu (Franky Weber Faust) e no time SQL Server o Fabiano Amorim, o Dirceu Resende e o Fabrício Lima.

Vocês podem se inscrever pelo link: https://www.sympla.com.br/happy-hour-com-dados-3—sql-server-vs-oracle__838710

A transmissão ao vivo será pelo YouTube e você precisa se inscrever no Sympla para receber o link do YouTube.

Evento: Happy Hour com Dados #3 - SQL Server vs Oracle

Espero vê-los lá para agregar com perguntas, comentários e fomentar a discussão.

Grande abraço,

Franky

Webinar: Como ter acesso a um Exadata para testes

Fri, 28 Feb 2020 08:40:20 -0800

Participe do webinar e descubra como ter acesso ao seu próprio banco de dados disponível a qualquer momento num Exadata para poder testar smart scan, hybrid columnar compression, storage indexes, entre outros recursos do Exadata. Vamos ver alguns casos práticos de smart scan, storage indexes e HCC. Testes ao vivo, tudo pode acontecer...

Evento ao vivo
Vagas limitadas
Quando: Terça-feira - 03/03/2020 19:30 BRST
Onde: https://gobrunch.com/events/39067
Realização: GUOB
Palestrante: Franky Weber Faust

Franky trabalha na Pythian como Consultor de Banco de Dados Oracle. Tem amplo conhecimento em Oracle Exadata e tecnologias de alta disponibilidade, também tem conhecimento em outros SGBDs como MySQL, Cassandra e SQL Server. Está sempre aprimorando suas habilidades focando principalmente no desempenho de Oracle Databases e Alta Disponibilidade. Já esteve envolvido em grandes implementações de Oracle RAC multi-nós em Linux, AIX e Exadata e também em implementações de ambientes multisite com Oracle DataGuard. Ele é OCP 12c, OCE SQL e Performance Tuning, OCA 11g, OCS Linux 6, RAC 12c, Autonomous Database, AWS Solutions Architect e foi nomeado Oracle ACE em 2017. É bastante conhecido na comunidade brasileira de banco de dados por seu blog http://loredata.com.br/blog e por todo o apoio e suporte na comunidade Oracle do Brasil. Franky escreve frequentemente para a OTN e é palestrante em alguns eventos sobre Oracle e banco de dados no Brasil e no exterior. Fique à vontade para entrar em contato através das redes sociais.

Espero vê-los lá. Abraços,

Franky

Online purging historical data efficiently using a single command in Oracle Database

Sun, 23 Feb 2020 07:17:38 -0800

Eventually, we are requested to delete some historical data from tables and if you are not aware of a few tricks you are probably in a huge mess. Deleting rows from a table generates a lot of undo and redo data. Undo data to support the rollback operation and redo data to guarantee the durability of your data and the undo data just generated for any recovery scenarios. When a lot of redo data is being generated means your database, that is expected to be in archive log mode, will generate a lot of archive log files. You might not be prepared for the amount of space required and so on... The space required for the archive log files will, obviously, depend on the amount of data you are deleting.

My point here is that, depending on your level of expertise, you probably know that, for example in a scenario you are keeping only 30% of the original data, you can CTAS just the portion of data you are keeping, drop the former table and rename it back to its original name. This way is much better and faster than a delete statement. I will not demonstrate that here as you can easily find articles on the web about that. If you do not know what I am talking about please read this one: Deleting many rows from a big table. Another option is dropping partitions if your table is partitioned, but partitioning requires Enterprise Edition and extra licensing. I am presenting here an operation that can be executed in any Oracle Database Enterprise Edition version 12.2 or above without the need of any extra license. Maybe by removing the "online" clause from the statement you can also run this in Standard Edition, but please check the licensing guide first and test.

First of all, let's connect to our database. Here I am using my Autonomous Transaction Processing Database TESTDB:

Let us check the size of the table we are working with:

A...Read More

Change NTP server on all Exadata nodes (<=18.x)

Tue, 11 Feb 2020 02:44:06 -0800

Once in a life time we might have to change the NTP servers associated with our Exadata systems. The Exadata servers run Linux and like any other Linux system the change is pretty much the same considering you are not yet running the Exadata image 19.x or higher since it comes with Oracle Enterprise Linux 7.x which uses chronyd instead of ntpd. So the procedure presented here is for any Exadata running 18.x image or lower and applies to OEL 6.x. for the ntpd service.

Even though I'm presenting here how to change the NTP configuration, you can use the same approach for basically any file you might have to change on all nodes of your Exadata system.

When we are working with Exadata it is nice (but not secure) to have passwordless SSH configured to enable us to run commands in parallel on all nodes without prompting for password for each node. We can test if passwordless SSH is configured just by using dcli to open a SSH connection to each node of our all_group file:

all_group file contains one row for every node we have in our Exadata system, in my case here it is an Elastic Rack:

15 Storage Servers
6 Database Servers
3 Infiniband Switches

Saying that the hostname command executed worked well we are ready to move forward and validate the current NTP configuration by querying the configured NTP servers from the ntp.conf file on all nodes:

We have identified the rows we have to change in the file. Before performing any change let's test if the current and new NTP servers are responding from these Exadata nodes using the command:

Considering everything is working as expected we can now backup the ntp.conf file in case we need to rollback our change:

OK, we are finally ready to change the NTP configuration. We are using sed here to ease our life so we don't need to go to any text editor like vim to open the ntp.conf file on each node. It is up to you how you prefer performing this change. I...Read More

Standalone Server com Oracle GI 12c, passos iniciais

Fri, 29 Jan 2016 09:15:56 -0800

Olá pessoal, depois de termos instalado o Grid Infrastructure 12c para nosso Standalone Server e o Oracle Database 12c, criamos o banco de dados, agora precisamos saber como administrar este ambiente.

Vamos ver alguns comandos para realizar tarefas administrativas no nosso Standalone Server.

Para administrar o Oracle High Availability Services num Standalone Server você deve estar conectado como root. O HAS é um grupo de serviços responsável por identificar falhas, iniciar, parar e reportar erros do ambiente. Iniciando ou parando o HAS todos os demais serviços dependentes dele são consequentemente afetados. Se você efetuar um shutdown do servidor Linux todo o processo de parada do HAS irá ocorrer automaticamente e iniciará da mesma forma no boot.

CRSCTL

Caso queira desabilitar o HAS para não iniciar automaticamente no boot do Standalone Server:

Para habilitar novamente:

Desabilitar não para os serviços, da mesma forma que habilitar também não inicia.

Para verificar os recursos que estão rodando no servidor e também saber o status deles:

Para verificar um recurso específico:

A opção -t mostra a saída em formato tabelado:

Não é possível parar os recursos com o comando crsctl:

SRVCTL

O tipo e o nome do recurso devem ser passados para que o comando funcione:

Somente o nome do banco de dados e não do recurso:

O comando acima faz um “shutdown immediate” conforme é possível comprovar no alert_loredb.log:

Da mesma forma que o...Read More