Systems Alliance Blog

Server virtualization through various hypervisor technologies is prevalent in every data center I visit.  There may be one out there somewhere not virtualizing servers, but I haven't seen it.  The efficiencies are just too magnificent to pass up.  Power, cooling, cash, personnel, and floor space all make the list of significant cost savings gained through server virtualization.

What about storage virtualization?  Will you get the same economies virtualizing storage as servers?  Let's examine a few fundamentals first.  Virtualizing a server essentially means  running multiple virtual servers on one physical server, presumably with each one performing different functions.  However, with storage, we're already doing that.  We take our storage subsystem, carve it up into pieces (LUNs, for example) and give it to servers for various uses.  So what is storage virtualization?  My own simple definition is this:  If you can manage and provision all of your individual storage systems (Sun, EMC, IBM, HDS, etc.) from one “pane of glass,” and seamlessly move your data between different systems, then it’s virtualized.  I found a fancier definition on Tech Target, which reads: “Virtualization is the pooling of physical storage from multiple network storage devices into what appears to be a single storage device that is managed from a central console.”

I believe there's a consensus that storage is next realm of the data center to make the virtualization jump.  Many vendors are starting to offer storage virtualization solutions. FalconStor is working with VMware to combine the virtual server and virtual storage worlds (read http://www.falconstor.com/en/pages/?pn=VMware).  If you need to move data from one array to another, that is a simple task in the virtual environment.  Storage administrators now have the option of provisioning storage from a pool of storage types, offering the right type of storage to the right applications.

In a virtual storage environment, data protection becomes a simple task.  Most storage virtualization vendors provide functionality for in-system replication (snapshots), replication (e.g., for disaster recovery), and thin provisioning, within their core virtualization platform. 

Using FalconStor again as an example, their Network Storage Server (NSS) product provides a virtual storage environment, thin provisioning, and DR replication in a single, low-cost appliance.  This functionality not only simplifies storage management, it creates huge potential cost savings. Because it over-rides and extends the functionality of your storage vendor’s proprietary management software, you no longer need to pay for ongoing licensing and support costs. If you've got a heterogeneous storage environment – and multiple sites – the savings just multiply.

I believe as more IT professionals realize the cost savings, flexibility, and ease of management of virtualized storage, it will become the next hot trend in moving towards a completely virtualized data center.

If you'd like to understand how you can benefit from storage virtualization, click here and request a no-obligation (free) white-board assessment session.

How well does your SAN handle crunch time? I’m talking about periods of heavy random read/write scenarios, such as:

• Data mining
• High Performance computing
• ERP Application loads

Depending on how you’ve got things architected, you should have no problem provisioning high-performance storage for all your applications. Got a slow SATA array? You can greatly boost its performance by applying cache principles to the entire SAN, not just a single array – and, not just a write cache, but an intelligent read cache. 

Now, imagine taking this a step further. Let’s add an SSD cache in front of your entire SAN (operating at over 1 million IOPS) and make it intelligent with data profiling… OK, let’s turn on the lights and get real. You can make this happen today with the FalconStor NSS SAN Accelerator. NSS lets you configure a write cache and a read cache for your entire SAN using two technologies: SafeCache and HotZone.  Here’s how they work:

SafeCache is a designated area on your SAN, such as an SSD array, where all application writes are sent. By using an SSD array (or any faster LUN on your SAN), the writes are high performance, and acknowledgements are much lower-latency.  Writes are also sequentially written to the final SAN target, which is any typical SAN resource (SATA array, FC array, etc.)   You can see how this would greatly increase the overall performance of the SAN, and subsequently all applications utilizing the SAN storage. With all writes going to the highest performing LUN on your SAN, the write performance becomes predictable regardless of which array is the final target, and predictability is a very good thing in our business.

HotZone is also used as a cache, but it is a designated read cache. Unlike SafeCache, HotZone can be used for the entire SAN or prioritized by application. HotZone intelligently recognizes data that is accessed on a more regular basis, and makes a copy available on the SSD array (or other faster LUN you designate).  As the data is used less, the data profiler will remove the data from the HotZone and redirect access to its normal location on the SAN resources.  Again, you get faster, more predictable storage performance.

How fast you ask?
Lab tests of the FalconStor NSS SAN Accelerator  show that adding a solid-state layer of one percent of a SAN´s total storage capacity will increase overall SAN performance more than 100 percent. Compared with achieving an equivalent performance increase by adding spindles alone, the FalconStor NSS is a bargain – even for organizations with modest budgets.

If you’re having performance issues or contemplating adding disk to boost SAN performance or reliability, please contact us for a no-obligation storage assessment.