Clustering, going beyond the usual distributed storage arrangement
One of the recent developments in storage systems has been to
cluster distributed storage systems in an arrangement that presents a single image from a management standpoint but allows scaling in both capacity
When you register for SearchStorage.com, you’ll also receive targeted emails from my team of award-winning editorial writers. Our goal is to keep you informed on the hottest topics, the latest news and the biggest challenges you face as a storage professional today.
Rich Castagna, Editorial Director
and performance by adding additional controller elements. The clustering goes beyond the traditional dual controller arrangement seen in most distributed storage systems as depicted in Figure 1. This is not to be confused with the cache centric storage systems that typically have both mainframe and open systems connectivity with a centralized cache to which all interfaces have direct access.
Figure 1:
There have been a few products released that provide a single image storage system with clustered controllers. Notably 3PAR has its
InServ system which has been available directly from 3Par and recently XIOtech has released a clustered product, the
Magnitude 3D. Other mainline storage vendors are likely to follow suit. Either completely new products might be developed, or more likely, an existing distributed storage product with additional embedded code and hardware to enable the clustering will be offered.
The clustering of distributed storage systems is a major development for storage systems. The scaling can be near linear from a performance standpoint up to the implementation limit of the number of controllers. The real value is in presenting a single image from a management standpoint. From an attached server standpoint, the storage system presents LUNs and supports LUN management in some form so the clustering is relatively transparent. The value in management is important in a cost of ownership perspective in that it is much easier to manage one larger system (figuratively) than many smaller ones. This also allows a clustered distributed system to be used as a storage consolidation solution. The fact that it can scale in performance and capacity non-disruptively gives customers additional alternatives.
The number of controllers that can be used in the cluster, how failover and cache mirroring is handled between them, and the mechanisms to manage the clustering (both hardware and software) are going to be implemented differently by the different vendors. There will be quite a bit of differentiation for products from the vendors and customers will have to evaluate carefully the solutions to determine what products fit the requirements.
The early companies that have delivered products represented by 3Par and now XIOtech will be demonstrating the value of clustered distributed storage. We can expect other solutions to follow.
Dig Deeper
-
People who read this also read...
This was first published in January 2004