Christophe's Corner

With its V-Max announcement a while back and its recent earnings call, EMC is making noise about so-called federated storage, scale out architecture… and selling futures in a big way. Like any marketing term, federated storage has been heavy on implication and light on definition. Some of us believe, in fact, that federated storage is vitally important and Hitachi is already delivering on the vision not EMC, or any other vendor for that matter.

Federated Storage Defined
The folks at Wikibon recently published the following definition of federated storage:

Federated storage is the collection of autonomous storage resources governed by a common management system that provides rules about how data is stored, managed and migrated throughout the storage network. In this definition, storage resources include disk capacity managed by controllers or appliances controlling multiple arrays.

One core requirement implied by this definition is the ability to move application data within this federation without any user disruption whatsoever. The business driver here is that disruptions are very difficult to schedule and require enormous amounts of planning, often as much as six months. You’ll hear more on this topic in a future blog.

Tiered storage addresses this need well today, but it can get even better. While tiering optimizes storage efficiency within an array, when storage must be completely moved to another point of control (e.g. moved to a separate controller or appliance) tiered storage falls short because the move is disruptive to applications (although our technology allows customers to mitigate this significantly, to minutes only in some cases).

How Disruptive is Non-disruptive?
Many vendors will claim they can move data non-disruptively. As a customer you need to carefully read the fine print. For example, IBM’s marketing people will claim you can move data from one SVC to another non-disruptively. The fact is you can do this within an SVC cluster but not across clusters.

EMC is also fond of invoking the term non-disruptive, when really they mean less disruptive. Or in the case of migrations from DMX to V-Max, hugely disruptive– unless you think forklift upgrades are minor events.

The truth is Hitachi Data Systems is the only block-based storage vendor today that can offer federated storage and achieve truly non-disruptive migrations for open systems, even across generations of arrays.

How can Hitachi make this claim?
Hitachi offers High Availability Manager which is a software feature that allows data controlled by one USP V to be replicated/migrated to a completely separate array. Once the migration has been completed, the application can access the new target array with no reliance on the original system. You can decide to “fail over” when you want. One more thing that is actually very important to many customers: we can support asset replacement and retirement of OEM external storage assets without “any” disruption to the customers application.

Please check our video on HDS TV here.

Can EMC do this today? When will EMC be able to deliver truly non-disruptive migrations? Your guess is as good as mine.

As a customer, you have choices:

1. Buy functionalities today that meet your tactical and strategic requirements for federated storage, achieve fast ROI and future-proof investments, or…
2. Buy futures from vendors that have consistently forced you to endure expensive, time consuming and disruptive migrations when there are better options available.

When you think about it in these terms, the choice is pretty obvious.