Storage Virtualization Showdown (NetApp, IBM, EMC and HDS)

Storage virtualization isn’t new, it has been around in one form or another since the invention of the filesystem.  The focus of this series, however is on storage virtualization within the SAN fabric.  The primary requirement for a vendor to be included in this “showdown” is the ability to virtualize 3rd party storage systems.  I’ve also limited the list to what seems to be the most considered products available today.  If your product was not included or if you feel like it was misrepresented, feel free to whine in the comments section.  I will update the series to reflect anything that is factually incorrect.

Why Virtualize?

Good, storage virtualization can significantly improve many aspects of a storage environment from both a technology and business perspective.  The agenda for the series is as listed below.

Each of these areas are key to solving storage challenges common in many mid-to-large enterprises.  I’ll dig more into how in each of the posts within the series.  For now, let’s take a look at the contenders.

IBM SVCimage

First Ship: June 2003
History: Based on COMPASS (Commodity Parts Storage System) project developed in the IBM Almaden Research Center
Hardware: X86 Architecture
Install Base: 21,000+ Nodes, 6,900+ Clusters

HDS VSPimage

First Ship: Sept 2010
History: Based on USP-V and previous USP architectures
Hardware: Monolithic and x86 Hybrid Architecture
Install Base: 1,400+ VSPs, 12,400+ Combined USP-V/USP-VM/VSP

EMC VPLEXimage

First Ship: May 2010
History: No Predecessor (rumored to have technology sourced from both Yotta Yotta and VMAX) 
Hardware: x86 Architecture /w Proprietary EMC Interconnects
Install Base: Unknown

NetApp V-Seriesimage

First Ship: 2003
History: Based on existing FAS Architecture, originally shipped as HDS branded “gFiler” and was later productized by NetApp as V-Series
Hardware: x86 Architecture
Install Base: 5,000+

First Look…

The easiest way to get a quick understanding of the capabilities of the hardware is through capabilities scorecards.  However, don’t assume that having a better “score” in the matrix makes for a better product.  The best way to read matrixes like the below are to ensure the products you are considering meet the flexibility and scale needs of your environment.  For example, if you only need 32 storage ports, a maximums difference between 64 and 192 is irrelevant.  Also keep in mind that most environments will have multiple storage virtualization systems, which allows the below numbers to scale horizontally as well.  I’ve highlighted platforms that offer the most options or scale in each area to make the chart easier to read.

image

There are a few high-level architecture differentiators that are worth noting, and it is why I’ve included this chart. 

HDS VSP is a clear leader in terms of high-end scale.  This is largely due to the monolithic and x86 hybrid architecture that has its roots in the USP and USP-V predecessors.  However, scale of this magnitude may only really matter in the largest enterprises. VSP is also the only storage virtualization platform to allow FICON support for Mainframe connectivity. This can be an important consideration if you plan on virtualizing your Mainframe storage.

NetApp V-Series offers the most flexibility in terms of open systems connectivity with both block and file protocols available right out of the box.  Other vendors only support FC and/or iSCSI protocols and require a NAS gateway to enable file connectivity.  While the VSP supports the greatest SDRAM traditional cache, the V-Series supports the greatest total amount of cache by allowing SSDs to be used as a 2nd level cache tier.

That’s it for this intro.  There is much still to consider and as the series continues, I’ll dig more into what differentiates the above platforms in the areas that are most important to enterprises looking to virtualize. 

Leave a Reply

 

 

 

You can use these HTML tags

<a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>