What is the Evolution of Storage Systems?

Are you ready to learn about hyper-converged infrastructure? Do you want to know why we removed storage from servers and then put it back? If so, keep reading, and let's get into it. 

What is the Evolution of Storage Systems? 

Storage solutions in the enterprise IT world went through three primary evolutions. If you’re new to IT, you might recognize some of these solutions for home-based products like UnRaid or TrueNas. Let’s walk through each evolutionary step and why each differs.

An Overview of How Storage Has Evolved

In this video, CBT Nuggets trainer Jeff Kish covers the evolution of enterprise IT storage systems. He explains how the IT world evolved from mounting storage solutions in the server to making them external in the SAN — and then re-integrating storage through hyper-convergence.

Servers with Storage Built-in

The server environment was static for a long time. Both servers and personal computers came with built-in storage. There weren’t any other options. Let’s explain. 

Servers need access to storage as fast as possible. They typically work with data on demand. For example, a website might need a database, which needs to be accessed constantly. 

At the time, you could move storage away from the server through an external hard drive or a network-attached storage unit. These types of devices are slow, however. They are best used for archiving or storing cold copies of data. They aren’t fast enough to work with data on demand. 

Keeping storage on the server created inefficiencies. Each server needs to be managed individually, which means configuring things like storage quotas, access policies, etc. for each device. 

Predicting how much storage each server needs can be complicated. Unlike a SAN or HCI, servers with built-in storage only have access to the internal storage. While one server might only utilize 10% of its storage space, it can’t easily share that storage with another server using 90% of its storage. 

Servers with Storage with Converge Infrastructure

Around early 2010, companies like Cisco and HP started working on a converged infrastructure. Storage was removed from servers and integrated into a SAN. While this made managing storage solutions easier, a converged infrastructure didn’t come without its complications.

The primary reason for those complications is that a converged infrastructure typically isn’t more than a wrapper for support around storage systems. Of course, this makes supporting these converged infrastructures far easier, but the SAN is still made from various components. 

This isn’t to say that converged infrastructures for storage systems didn’t have benefits. A SAN is far easier to manage. Administrators don’t need to configure multiple devices, and SANs can be managed through the same management interface. 

SANs can utilize arrays, various management apps, and better file systems. This isn’t to say that a server can’t use something like ZFS or a RAID config, but utilizing these technologies in a server becomes much more expensive at scale. For instance, a server using RAID 5 requires at least three drives per server. On the other hand, a SAN can keep scaling its RAID 5 array without requiring additional drives for parity. 

Servers with Storage with Hyper-Converged Infrastructure

Finally, we have hyper-converged infrastructure (HCI). A system that uses HCI moves storage drives from the SAN back into the server. However, there’s a difference between HCI and the original built-in solution.

In a hyper-converged infrastructure, storage solutions are managed by software. That software is closely tied to the hardware. If you understand anything about software-defined networking, you’re on the right track for understanding a hyper-converged environment. 

I will oversimplify this as a warning to you IT gurus, so don’t cringe too much. Let’s say we have three ‘servers’ in our hyper-converged environment. Server 1 has three drives, Server 2 has 20 drives, and Server 3 has ten drives. 

The HCI software combines all of those drives into one large storage volume. Various pools are then created from those volumes. This is done from a management application. 

In short, each server sees one large pool of storage. Data can be shared between servers; administrators only need to interact with that management application to configure each server. 

The best example of that management application is Cisco UCS. When you hear' UCS, ' you might typically think of networking solutions, but Cisco Unified Connect Software can also manage storage systems like it can a software-defined network. 

Get Trained on Hyper-Converged Infrastructures and Storage Options

Are you interested in learning more about hyper-converged infrastructures and storage solutions? In that case, look at CBT Nugget’s Cisco Hyperflex online training. Get access to some of the best training modules at your convenience.