Originating Author: Dave Vellante
One of the most common flaws of storage total cost of ownership (TCO) is comparing two configurations or storage approaches that appear similar but are actually quite different. This can lead to erroneous conclusions and poor technology choices. Beware of blanket vendor claims declaring one technology is superior to another.
Horses for courses
Take the example the impact of workload on manageability. Some call this “horses for courses?” It’s a phrase from horse racing that means a horse may run well on certain racetracks but not so well on others (for example a big ranging colt might like the wide sweeping turns of Belmont Park but not fare so well on the tight turns at Pimlico). Taken in an IT applications context, the workloads you deploy for a specific storage infrastructure will have a major impact on the cost and performance of the system.
Buyer beware
This seems obvious but I’m sometimes stunned by the lack of attention paid to the applications when statements are made and research cited that definitively declares things like “NAS is cheaper to manage than SAN.” One of the most important things to consider in any choice of infrastructure is: “What work is being done by the system and how is this affecting the projected benefits?” As an example, consider storage management productivity, a common indicator of staff cost. Otherwise known as “Gigabytes Managed Per Person,” this metric tracks the amount of storage managed per full time equivalent (FTE) and directly influences IT staff expenses. The examples below show two scenarios that reflect three-year TCO based on storage management productivity. Each scenario compares:
- A NAS-based infrastructure with twenty (20) Windows servers. The NAS box is configured with 6TB of protected storage accessed by 2,000 users;
- A SAN with advanced backup and copy services supporting twenty (20) of the same class servers. The SAN is also configured with 6TB of protected storage accessed by 2,000 users.
3 Year Staff Costs:
Scenario I:
- NAS: $167K
- SAN: $214K
Scenario II:
- NAS: $300K
- SAN: $250K
Know thy workload
The conclusion drawn from scenario I: “NAS is cheaper to manage than SAN.” The conclusion in Scenario II: “SAN is cheaper to manage than NAS.” So, what’s different?
The catch is workload. Scenario I measures a predominantly file/print workload on the NAS infrastructure while Scenario II measures a transaction-oriented workload on the SAN (primarily business processing). Scenario II shows the same configuration but both NAS and SAN are running the same mixed workloads.
In scenario I, the NAS is the better choice because the added expense of SAN is not warranted for the file/print workloads. In scenario II, the NAS infrastructure has to be over-configured to accommodate the challenges of supporting more SAN-friendly workloads-- SAN is the better choice here.
Bottom Line: The complexity of the workload will have a major impact on overall storage costs and potential for efficiency improements. In general, workloads that are more complex are more expensive to manage and will deliver greater cost payback when consolidated. -Dave Vellante
What more on this topic? Check out Consolidating storage