Freely subscribe to our NEWSLETTER

So what is Clustered Storage?

Modern storage clustering has gone way past the simple fail over between a pair of redundant server/controller head and disks. Today, most advanced systems tend to be Distributed Clustered Storage. This architecture is a networked storage system that allows users to add self-contained nodes to continually expand the cluster. Each node contains processing power, cache, terabytes of disk storage and front and backend interconnect normally based on Gigabit Ethernet, InfiniBand or Fibre Channel.

Each node also has a suite of built-in applications to deliver the three layers of traditional storage architectures; file system, volume manager and RAID. This creates an intelligent fully symmetrical file system that spans all nodes within a cluster. This allows each node to serve any data file to any client irrespective of where the data physically resides on the cluster.

Strengths:
Ability to scale performance and capacity independently
High availability able to remain operational even with multiple simultaneous drive or node failures
Highly automated management functions including load balancing, drive rebuilds, data replications.

Weaknesses:
No independent interoperability standards between vendors
Not suited to highly structured “small” files like email or database transactions
Unable to virtualise existing NAS or SAN hardware
So what is Storage Virtualisation?

Storage virtualisation is essentially an aggregation technology that presents a single “virtual” storage infrastructure normally derived from different storage end points. These solutions vary greatly and are often a combination of software applications, appliances and switches to create a single namespace of storage that appears to management systems as one large pool of data. Typically, these solutions enable “synthetic trees” that encompass several NAS servers or storage devices. Most virtualisation solutions can control laying out a file (striping data) across disk volumes to a specific silo but often not across the silos that make up the virtualisation. The ability to categorise data makes virtualisation suitable for information lifecycle projects, as it will often allow data movement between tiers of storage with limited client interruption.

Strengths:
Ability to tie together multiple storage vendor products under a single virtual namespace
Prolongs the life of otherwise redundant storage platforms
Reduces management overhead compared to traditional “islands” of storage

Weaknesses:
Limited scalability for both capacity and performance
Limitation on largest file size and largest single name space imposed by underlying NAS silos
Increased complexity especially during drive or controller failures

So is there a winner?

Both technologies aim to reduce the cost of managing data and succeed in many respects. As an analogy, Clustered Storage is building an infrastructure from the ground up to deal with large files and huge volumes of data. Where as Storage Virtualisation is a short-term response to the problem of spiralling data management using tools to help alleviate the burden until organisations can take a pause and create a lasting fix.

The factors involved in each potential customer’s circumstances are too varied to make a clear-cut “winner” but a simple rule can be extrapolated for the casual observer. If an organisation wants to build a multi-terabyte capacity storage pool suitable for large capacity data files then clustered storage is a better option. If an organisation has invested in a lot of NAS and SAN hardware that is complex and expensive to manage then Storage Virtualisation offers the most immediate benefits. Depending on the situation, each technology can allow the customer to significantly improve their ability to manage data storage.