Oracle Cloud Infrastructure (OCI) Fundamentals Part 3 – Storage Services….

In this installment, we shall look at the storage services that are available with Oracle Cloud Infrastructure. Let’s get started.

In the OCI Storage services, there are various types of storage options available. Of course, all of them offer different benefits and are meant for different use cases. Before explaining them, here is the list,

  • Block Volume
  • Local NVME
  • File Storage
  • Object Storage

Since it’s wide variety of options, it’s important to remember that not every storage type is meant for storing similar type of the data. Also, besides the type of the data, other factors also play an important role while making the choice. For example, whether the storage is going to be a permanent storage or not? What’s the workload type that you are expecting and also your expectations from the performance delivered by the storage. Also, since performance is mentioned, how the storage is going to be accessed. For example, you can have a storage available locally or it can be a DAS(Direct Attached Storage) or it can be a remote storage. Thus, all of these factors and some more, may govern that which storage will fit in your requirements.

Block Storage

Block storage is like using a normal storage except that storage here is not available locally but will be provisioned from a remote chasis. Since it’s a formatted drive, it’s available by the operating system being a mounted drive. And it’s a block storage which means that you need to use some sort of file system managed by your operating system of choice e.g. NTFS if you are on Windows.

So, what is a block volume service? Well, it’s a storage service for the OCI compute instances. As mentioned, it’s storage only but available to the OCI compute instances via remote servers. So there is a separate server from your OCI instance and from it, the storage is provided. This storage service is of two types- Boot volume and block volume. Boot volume stores the operating system. And block volume stores the actual data.Now since these two types of the volumes are available as remote, they are independent from your compute instance in the terms of availability. It means that if for some reason, your compute instance dies, your disks for both Block and Boot will be available. Thus, this arrangement is very well usable in storing databases and also in public clouds.

To ensure that the block volumes offer high availability, block volumes store data in 3 separated Failure domains within one AD. Thus obviously, if one FD becomes unavailable for some reason, another 2 FDs will be still serving the data. But we must remember, Block volumes are still disks with the exception of being available remotely. That means you can still have issues like data corruption or user-induced wrong deletions of the data etc. Thus, just like in on-prem environments, you need to configure backups of your block volumes and even OCI allows you to have automated scheduled backups option. So the volum back will be stores in the object storage (we shall discuss it later). Now from this backup availale from this object storage, you can make another compute instance by creating a new volume. Of course, if you are having more than 1 AD in region, you can have the backup used in other AD’s. There are different tiers available for Block volumes differing in speed and also in the available space. So do choose your option after a careful review.

Local NVMEs

Well it’s also similar to block volume but it’s attached to the local compute instance and acts as a temporary storage. Of course, it’s very high in performance so if you are planning to have workloads which need high performance, or you are managing a data warehouse or NoSQL databases, this kind of disks performs very well. This storage does survive reboots but it serves for non-persistent data.

File Storage

File storage allows you to have a shared storage for your files using NFS among multiple compute instances. File storage allows to have backups created using snapshots. Security of the data is guaranteed via encryption. You can use File Storage in EBS, Big data etc. Just like data, File storage is also created in Fault Domains within an AD. Thus, just like block volumes, FD also get the data replicated in 3 different FD’s and also get the data encrypted.

Object Storage

Object storage maintains storage of the data in the form of objects and each object is stored in a bucket. Each object maintained by Object storage, comprises of object metadata and the original obbject too. Now unlike a compute instance, object storage service is provided to a region. You can use this object storage service for storing archives (or backups too), log data from text files etc. The way the objects are stores in the object storage is very different from the normal file system. In a normal file system, to manage files, there is the concept of folders. But in an object storage, there is no such folder concept and each object is stored in a single, flat structure. Of course, this improves the performance dramatically. Object storage is accessed through HTTP methods like GET, PUT etc. So where do you use an object storage? Well it can be used to store non-structured data i.e. images, movies etc. Also, object storage can be used to store archives or backups, log files data for troubleshooting or auditing etc. Additionally, object storage is not tied to any specific compute instance but is offered as a regional service.

There are two tiers in which Object storage service is offered- standard and archive. As the name suggests, standard makes the objects accessible immediately. Also, standard storage tier serves the latest copy of the data always and the data retrieval is immediate. On the contrary, archive tier is meant for that data which isn’t accessed often yet it’s needed to be preserved e.g. auditing data. If you are using archive tier, data retrieval can take longer-first byte retrieval can up to 4 hours. Also, this tier is significantly cheaper than the standard tier. So if you are planning to access the data immediately, choosing archive tier of object storage isn’t going to help much.     

So this concludes a rather very simplified overview of the storage service offered by OCI. Stay connected for the next installment.

Hope it helped.