October 8, 2021 By Drew Thorstensen
Anna Matetic
2 min read

Instance storage on IBM Virtual Servers for VPC is an excellent option for fast, affordable, temporary storage that is widely used among developers for their cloud-native workloads and services. 

Instance storage is a storage technology that is complementary to, and can be used alongside, VPC’s block storage volumes. 


Block storage boot and data volume lifecycles persist independently from the virtual server instance to which they might be attached. By contrast, the lifecycle of instance storage is tied directly to the lifecycle of the virtual server instance with which it was provisioned. 

When the virtual server is shut down or deleted, the instance storage attached to the virtual server instance is cryptographically erased, and the instance storage is no longer available. The virtual server’s boot and data volumes are not affected. The temporary nature of instance storage makes it an ideal storage option for Distributed File Systems, Content Delivery Networks (CDNs) and temporary files created during transaction processing. 


The performance of instance storage exceeds that of block storage. It scales linearly based on the amount of storage consumed, thereby providing a consistent performance profile. This scaling is determined by the amount of storage that is provided to the instance. 

By using Quality of Service, the IOPs and bandwidth are scaled based on the amount of disk that is used. Rate limits are placed individually on each disk. If you use multiple disks, speeds might be improved by striping the data across the disks.

Instance storage is available in the Balance, Compute, Memory, Very High Memory and Ultra High Memory virtual server and dedicated host profile families. You can expect the following approximate performance results:

  • Read IOPs: ~200 IOPs per GiB, up to approximately 300k IOPs total
  • Write IOPs: ~65 IOPs per GiB, up to approximately 75 IOPs total
  • Read bandwidth: 0.85 MiB/GiB
  • Write bandwidth: 0.85 MiB/GiB

FIO performance benchmark example

FIO benchmarks were run against several different Balance family profile types with instance storage. The tests were run with a simple synthetic benchmark balancing the I/O across all disks, with the following results. As you can see, the scaling of the virtual service instances are linear until the caps are observed on the IOPs:

Instance storage FIO performance benchmark example.

Most workloads don’t have a unified I/O pattern and mix random small block operations with sequential large block operations. As such, your workload might see different results that are based on its unique I/O pattern. Therefore, your performance might vary depending on the characteristics of your workload. 

With a solid understanding of the lifecycle and performance characteristics of the disk, instance storage on IBM Virtual Servers for Virtual Private Cloud (VPC) can be a helpful component to improve the performance of your cloud workloads.

Learn more

More from Cloud

Sensors, signals and synergy: Enhancing Downer’s data exploration with IBM

3 min read - In the realm of urban transportation, precision is pivotal. Downer, a leading provider of integrated services in Australia and New Zealand, considers itself a guardian of the elaborate transportation matrix, and it continually seeks to enhance its operational efficiency. With over 200 trains and a multitude of sensors, Downer has accumulated a vast amount of data. While Downer regularly uncovers actionable insights from their data, their partnership with IBM® Client Engineering aimed to explore the additional potential of this vast dataset,…

Best practices for hybrid cloud banking applications secure and compliant deployment across IBM Cloud and Satellite

10 min read - Financial Services clients are increasingly looking to modernize their applications. This includes modernization of code development and maintenance (helping with scarce skills and allowing innovation and new technologies required by end users) as well as improvement of deployment and operations, using agile techniques and DevSecOps. As part of their modernization journey, clients want to have flexibility to determine what is the best “fit for purpose” deployment location for their applications. This may be in any of the environments that Hybrid…

Level up your Kafka applications with schemas

4 min read - Apache Kafka is a well-known open-source event store and stream processing platform and has grown to become the de facto standard for data streaming. In this article, developer Michael Burgess provides an insight into the concept of schemas and schema management as a way to add value to your event-driven applications on the fully managed Kafka service, IBM Event Streams on IBM Cloud®. What is a schema? A schema describes the structure of data. For example: A simple Java class…

SSD vs. NVMe: What’s the difference?

7 min read - Recent technological advancements in data storage have prompted businesses and consumers to move away from traditional hard disk drives (HDDs) towards faster, lower-latency solid-state drive (SSD) technology. In this post, we’re going to look at this new technology, as well as the fastest and most popular protocol available to connect it to a computer’s motherboard—non-volatile memory express (NVMe). While the terms SSD and NVMe are often used to describe two different types of drives, they are actually different data storage…

IBM Newsletters

Get our newsletters and topic updates that deliver the latest thought leadership and insights on emerging trends.
Subscribe now More newsletters