A capable, easy-to-use cloud product
December 15, 2020
A capable, easy-to-use cloud product
Score 7 out of 10
Vetted Review
Verified User
Software Version
IBM Cloud Bare Metal Servers - VPC
Overall Satisfaction with IBM Cloud Bare Metal Servers
We use IBM Cloud [Virtual Private Cloud (VPC)] in our team for dynamically scaling our compute capacity as the application needs it. As our application's demands change significantly on a minute-to-minute basis, we've found it much simpler and cheaper to use IBM Cloud to dynamically provision resources than to reserve the application's maximum needed capacity in our internal IT infrastructure.
- Easy and fast provisioning of resources through UI, API and CLI
- Huge scalability
- Great customer service
- Viewing online invoices and billing data is awful - there are several separate systems that require multiple layers of accounts
- Not all features are available in all regions, such as single-site resiliency in COS
- VPC instance disk drives are awkwardly slow. The boot drive is limited to 50MiBs read/write, even though network is 2GiBs.
- Our dynamically-launched instances start faster than they did on AWS
- The connection to Cloud Functions has allowed us to get massive speed up by parallelizing IO- and CPU-intensive tasks
- Amazon Elastic Compute Cloud (EC2)
We chose IBM Cloud [Virtual Private Cloud (VPC)] because of its tight integration with the Lithops library, which allows us to quickly scale our compute resources by provisioning VPC instances or calling Cloud Functions, handling all the deployment, data moving, etc. without us needing to know the details or learn another language, all from Python code.
- IBM Cloud Functions
- IBM Cloud Object Storage
Cloud Functions in conjunction with Cloud Object Storage allows massive scalability in IO and CPU speed. Being able to read, processing and write 10s of GBs per second is pretty amazing, considering that you can scale up to 1000x4GB cloud functions in under 10s. There are several "embarrassingly parallelizable" parts of our data processing pipeline that have gone from 10s of minutes to 10s of seconds by rewriting them to use Cloud Functions.