Infrastructure: Converged (CI)

A widely used deployment architecture, optimized for cost-effective scalability and better hardware utilization.

Intro 

Nowadays, most IT infrastructures are fully virtualized. Two prominent infrastructure types are Converged Infrastructure (CI), where independent components including compute, storage, and networking are purchased from and integrated by a single vendor, and Hyperconverged Infrastructure (HCI), where compute, storage, and networking are merged into a single unified virtualization platform. Despite their differences, hyperconverged and converged infrastructure designs can coexist, as they address different needs. When you need to scale compute and storage resources independently, converged infrastructure is the way to go. 

Problem  

Organizations often need to expand compute and storage resources separately, particularly in large data centers. Hyperconverged infrastructure doesn’t easily accommodate this need. Scaling hyperconverged clusters means upgrading both compute and storage hardware simultaneously. Let’s break that down. Virtualized environments expand in two main dimensions: Compute (CPU, RAM, GPU) and Storage (flash, drives). 

In hyperconverged systems, each host runs both compute and storage layers on the same hardware. Thus, increasing one dimension means you must increase the other too, even if it’s not needed. This results in wasted money on unnecessary hardware. 

Solution 

Separation of data and compute isn’t the enemy when you have the right software. StarWind supports both hyperconverged and converged infrastructures. By separating storage and compute, you can scale these resources independently, which is more cost-effective. This allows your system to better fit your specific needs without wasting money on unneeded hardware. 

Using StarWind in a converged infrastructure means you can simply add more RAM or GPUs to a hypervisor node or add more hypervisor nodes to the cluster if you’re running out of compute power, without changing storage capacity or performance. Similarly, if you’re short on storage performance (IOPS) or capacity (TBs), you can upgrade storage without affecting compute power by adding a storage node. StarWind brings the benefits of converged infrastructure to the table. 

Conclusion 

StarWind can operate as Software-Defined Storage (StarWind VSAN) on your existing hardware or as ready nodes (StarWind Storage Appliance) in your environment. Running StarWind as a separate storage layer from the compute layer ensures that you only acquire the hardware resources you actually need, significantly optimizing both CapEx and OpEx.