Lesson 4: OCI Storage Services

Overview

Block Storage

• Like a hard drive in a server except the hard drive happens to be installed in a remote chassis. 如同服务器中的硬盘驱动器，但硬盘驱动器恰好安装在远程机箱中

• Data is typically stored on device in fixed sized blocks (e.g 512 Bytes).

• Accessed by operating system as mounted drive volume. Block storage 通过操作系统作为已安装的驱动器卷进行访问

• Applications/file systems decide how blocks are combined and accessed.

• Data is stored without any higher-level metadata e.g for data format, type or ownership. 数据存储时没有任何更高级别的元数据，例如数据格式，类型或所有权

• You can place any kind of file system on block level storage. E.g Windows uses NTFS; VMware uses VMFS.

Block Volume Service

• Storage for compute instances (REMOTE)

• 2 types of Block volumes:

  • Boot Volume (OS disk)

  • Block Volume (data disks)

• Why do we use Block Volume: it lets you store data independently and beyond the lifespan of compute instances. 它使可以独立存储数据和独立于实例的使用寿命

• When instance dies, Your boot volume and your data volumes are all still available. So Block Volume is used for data durability 数据持久化.

Block Volume Use ases

• Databases

• Exchange (supports block level storage only)

• VMware (VMFS volumes on block level storage)

• Server boot (in public clouds, instances are configured to boot from block level storage)

Block Volume Highly durable

• Storage is highly durable and persistent

• Block Volume store replica of data in 3 separate FDs

• You don't need to configure any software based protection (RAID 10 etc)

• A backup is recommend for minimize losing data.

Block Volume Backup

• Complete point-in-time snapshot copy of your block volumes

• Encrypted and stored in the Object Storage, and can be restored as new volumes to any AD within the same region

• Can copy block volume backups from one-region to another

• Backups can be scheduled

Block Volume Tiers

Local NVMe

• Temporary storage locally attached directly to the compute instances

• high-performance local storage

• Storage is non-persistent (survives reboot)

Local NVMe Use cases

• NoSQL DB

• in-memory DB

• On-Line Transaction Processing (OLTP, 在线交易处理)

• Scale-out transnational DB

• Data warehousing (数据仓库)

File Storage

• Hierarchical collection of documents

• Distributed files standards

  • NFS

  • SMB

File Storage Service (FSS)

• Shared file system storage for compute instances

• Support NFSv.3 distributed fils system

• Data protection: Snapshots

• Security: data-at-rest and in-transit encryption for all file systems & metadata

File Storage Highly durable

• Highly durable and Persistent

• File storage stores replica of data in 3 separate FDs

  

File Storage Uses cases

• Oracle Applications

• HPC

• Big Data and Analytics

• General purpose File systems (e.g. NAS)

Object Storage

• All data is managed as objects, for storing unlimited of unstructured data.

• Each object is composed of object itself and object's metadata

• Each object is stored in a bucket1, this means that accessing individual objects is fast and easy.

• Without a folder hierarchy

• High scalability and reliability

• Object storage relies on standard HTTP verbs (GET, PUT, POST, UPDATE, DELETE )

Object Storage Service

• Regional service, not tied to any specific compute instance

• Offers 2 distinct storage classes

  • Hot Storage

  • Cold Storage

Object Storage Tiers

1. Standard Storage Tier (Hot)

• Fast, immediate and frequent access.

• Data retrieval is instantaneous. 数据检索是即时的

• Always serves the most recent copy of data when retrieved. 检索时始终提供最新的数据副本

• Standard buckets can't be downgraded to archive storage. 无法将标准存储桶降级到归档存储

2. Archive Storage Tier (Cold)

• Rarely accessed data, must be retained and saved for long period of time.

• 10 times cheaper than Standard Storage Tier

• 90 days minimum for storage

• Objects need to be stored before download; Time To First Byte (TTFB) after restore request is made: 4 hours

• Archive buckets can't be downgraded to standard storage

Object Storage Highly durable

• Highly durable and Persistent

• File storage stores replica of data in 3 separate FDs

• In a multi-AD region, it stores replica of data in more than on AD

• Data detected and auto repaired

• For disaster recovery

Object Storage Use cases

• Content repository for data, images, logs and video etc.

• Archive/ Backup for longer periods of time.

• Storing log data for analysis and debugs/ troubleshooting.

• Storing large data sets (genome data, IoT).

• Big data, Hadoop

1 bucket: is a logical container for storing objects