What you will learn from this tip: Virtualization gives you the ability to utilize the total capacity of all of your storage systems despite their locations or dedication specifications. This tip explains why moving away from discrete storage systems will only save you management time and money in the long run.
The value of storage virtualization is so compelling that it will reinvent storage networking. Having discrete, individual storage systems to manage and support will one day be viewed as archaic and impractical. The customers that have implemented storage virtualization are seeing real value, reducing both cost and complexity in their environments. The main reason that people are implementing storage virtualization is to simplify management. Additionally, they want to reduce hardware and software costs. The Enterprise Strategy Group found that early adopters of storage virtualization on average saved 24% on hardware costs, 16% on software and 19% on SAN administration annually. If a customer had a budget of $1million, spending $500,000 on hardware, $200,000 on software and $300,000 on administration, the cost savings would be $209,000 per year.
There are NAS and SAN virtualization products that allow customers to consolidate management of multiple systems, implement intelligent tiered storage, extend the life of existing storage systems and centralize data protection.
To illustrate the impractical nature of discrete storage systems, let us examine a real world comparison of parking garages at an airport. There are four parking garages named A, B, C and D:
Garage A is the most expensive and closest to the terminal;
Garage B is just a bit further down and less expensive that Garage A;
Garage C is slightly less expensive than B and further away from the terminal;
Garage D is long term parking, which is much less expensive but requires a shuttle to get you to the terminal.
If the driver is in a hurry and absolutely requires the fastest access, then he or she will park in Garage A, even at a premium cost. If speed is not of the essence but cost is more important, then long term parking provides the best value. And for many drivers, Garage B and C offer a good compromise between convenience and price. Each driver can choose which garage makes the most sense for them. One day, a driver may need to use Garage A because they are in a hurry or are only going for a day trip and will return later that evening. Another time, that same driver may want to park in Garage D because he or she will be gone for over a week.
Imagine a situation that did not allow drivers to select the garage they could park in -- regardless of time constraints or cost issues each driver had no other choice but to park in a specific garage. Additionally, even if Garage A was nearly full, the designated vehicles have to wait for an open space instead of being re-directed to another garage. And even though the other garages have plenty of parking spaces, the decision is made to build an extension onto Garage A to create more spaces to accommodate the increasing number of designated vehicles.
The above example is very similar to the way that data storage is implemented. Customers may have multiple storage systems that are dedicated to a specific application or applications. And as that storage system fills up, it cannot share the resources of the other storage systems in the data center. Instead, more disk drives have to be added to it even though another storage system across the room may have a large amount of unused capacity.
Storage virtualization enables intelligent tiered storage, creating one large virtual storage system. Data can be moved transparently from one physical storage system to the other based on cost, capacity utilization and performance as needed.
So let's go back to the parking garage example. You park in Garage A because you are going a day trip. However, your plans change and you have to stay an extra week. Storage virtualization will move your car (data) from Garage A (expensive high end storage system) to the lower cost Garage D (low cost storage system). Additionally, when you get back your car will be exactly where you parked it in Garage A, but you will pay the lower cost (Garage D prices) for your extended stay. That is one of the core benefits of storage virtualization -- creating an intelligent tiered-storage environment that is transparent to users.
Storage virtualization doesn't just drop in and magically solve all of your problems. There has to be a great deal of analysis and planning. However, you can implement storage virtualization in stages. There are some customers who start out slow and incrementally implement. There are other customers that re-architect their entire storage network. And there are many points in between these two extreme approaches.
One of they key objections I hear concerning storage virtualization is that there is a single point of failure (SPOF) for all of your data. When evaluating any storage tool that is supporting mission-critical applications, you have to make sure that the system has no SPOF. Additionally, you might want to consider implementing disaster recovery for mission-critical applications. This is true whether you are using a storage system or a storage virtualization tool. Customers are consolidating multiple last-generation storage systems into next-generation storage systems that are bigger and faster. Any concern of SPOF applies to a single storage system that is being used for consolidation. Consider a storage virtualization tool as a way to consolidate storage systems, but with more options for creating a tiered-storage environment that could save you a ton of money.
Implementing storage virtualization doesn't have to be an all or nothing proposition. You can still have other storage systems that are independent. Also, think about having two storage virtualization products. They can even be from different vendors. That takes care of the other big objection to storage virtualization: Having a single vendor manage all of your storage.
For more information:
Virtualization delivers replication, mirroring
About the author: Tony Asaro is the senior analyst for Enterprise Strategy Group.
