sumetho - Fotolia
Block storage is used to provide low-latency networked storage that's used by databases and other applications. It requires a consistent network connection and doesn't do well with variability and delays. For this reason, cloud block storage is generally used with cloud-based servers and applications, rather than as a standalone storage service. Using cloud block storage as a back end for on-premises servers and applications isn't a typical use case.
The environment for block storage includes server instances that can read from and write to a block storage volume. That storage volume may be tied to a single instance or shared among several. The volume may be standalone or may be replicated elsewhere for data protection or for functional copy purposes such as analytics, creating new instances or some other application purpose.
When you're choosing block storage for cloud-based applications, there are six areas you should consider:
1. Persistent vs. ephemeral storage
Some cloud providers give you the option to use ephemeral storage, which stores data on the app server only until the cloud server instance shuts down, at which time the data is deleted. This is a good option for servers that run from a consistent image and don't need to write data to be saved for later usage. If it's a Web presentation server, for example, that receives data saved elsewhere and doesn't record any new transactions, ephemeral storage may be a simple and cost-effective solution. If it's a database server that will read and write, persistent storage that outlives any individual instance might be a better choice.
2. Storage media type
Like traditional on-premises storage, cloud block storage can be configured on multiple media types, with disk and solid-state drives (SSDs) being the most common options. Disk has more latency than SSDs, and can vary in performance depending on whether workloads are random or sequential, and on where a given piece of data resides on the spinning platter, but it is significantly cheaper. An SSD is faster for reads than writes, but both are faster and more consistent than they are on disk-based storage.
3. Guaranteed or best-effort IOPS
While most cloud services are sold as "best effort," some providers make it possible for you to specify and guarantee a given level of IOPS performance for block storage volumes, building in a form of quality of service (QoS). You pay more to guarantee the IOPS level that you need, but for applications that depend on consistent performance, this can make or break their cloud viability. Most guaranteed IOPS services require you to know the level you need, so this does require expertise and monitoring to ensure you have enough to avoid provisioning and paying for more than you need.
4. Throughput between cloud server and storage
The latency of the connection between a cloud server and its storage will have a big impact on overall application performance. Performance varies by provider, depending on how they build their data centers, servers and storage. There may also be variability to watch out for due to differences in one provider's facilities and architectures, so the more you can do to test configurations and loads, the better. Some providers have storage-optimized instance types that provide a faster connection to the block storage back end.
Cloud providers often describe the durability of their storage components in terms of annual failure rate. It's important to compare the durability of several providers, as well as the durability of your on-premises storage. The durability, combined with the level of component redundancy you configure, will yield your overall storage resiliency.
Block storage features include snapshots, replication tools, auditing capabilities, management tools, etc. These vary widely by provider. Make sure you understand all the bells and whistles that come with (or don't come with) a given cloud storage offering, as these will have a big impact on your ability to manage and optimize the storage resources you use.
Does durability lead to a significant increase in cloud availability?
The difference between object storage and file and block storage
Dig Deeper on Public cloud storage
Related Q&A from Andrew Reichman
With storage developments moving from hardware toward software, many vendors are reworking products to capitalize on the software-defined technology ... Continue Reading
When it comes to provisioning storage for virtual machines, techniques such as golden images and thin provisioning are key. Continue Reading
In physical environments data protection often means restoring an entire system, but using the same methodology for virtual servers is inefficient. Continue Reading