In his latest professional Alert, “Business Needs Dictate the Importance of Scale Up vs Scale Out Storage”, Consultant and Wikibon Analyst Scott Lowe makes an important point that is often lost in the discussion of new online architectures, Big Data, and other ‘sexy” subjects – while scale out is great for some applications, scale up is perfectly adequate and much less expensive for many others.

In storage terms, “scale out” means adding new nodes to a cluster. Each node comes with its own networking and controllers, and the cluster must have overall management that loosely couples the nodes so that they appear as one device. This is fine if the application set being supported requires massive amounts of storage – say in the Petabyte range.

However, many production applications – Lowe cites Exchange, Sharepoint and SQL databases as examples – do not require that scale of storage. When these applications outgrow their arrays, it is often much simpler and less expensive to scale up. Most arrays come built to accept expansion shelves. The advantage is that these shelves plug into the array and use the IO and controllers already in that array. Thus the cost per Gbyte of storage is lower, and the entire process is much less complex and puts minimum extra load on sometimes scarce data center resources.

Even some of the startup scale out storage vendors are recognizing the advantages of scale up in some use cases, Lowe writes. At Storage Field Day 4 in San Jose, Nimble Storage demonstrated its Scale to Fit architecture, which allows customers to choose the strategy that fits their need. And the latest systems from scale-out storage startup Nutanix also provide a scale-up capability.

So which is best to fix the problem of the moment? In most cases the answer is obvious, but when it is not the deciding factor is the extra IO and controller capacity in the existing array. If the application has outgrown those, then a scale out solution (or a replacement of the entire array with flash) is the answer. If not, then stick with the old fashioned scale up strategy. CIOs, Lowe writes, should choose storage that support both options to give them maximum flexibility to react to changing needs.

Graphic courtesy Scott Lowe