The cost of storage media is plummeting: both hard and solid state drive capacity has never been so inexpensive.
That would be great news for companies that have significant storage requirements, except for two inconvenient facts. Firstly, the purchase cost of a storage system is made up of much more than the cost of the storage media itself. And, more importantly, the purchase cost of a typical hard disk drive (HDD) based system accounts for less than 20 percent of the total cost of ownership (TCO), according to Parity Research.
So while the purchase price of a storage system is evident at the time that you hand over money to a vendor, it is not the whole story. What's important is the TCO over the lifetime of the system—typically about five years.
Some of the post-purchase costs are "hard" costs like power and cooling, which are easy to calculate up front. But there are also "soft" costs, which are more difficult to pin down and which often get overlooked. These soft costs include the cost of taking on risk—such as the risk of downtime or, less obviously, the risk of non-compliance with regulatory requirements.
"I think that a lot of folks that buy storage dramatically underestimate these soft costs, particularly the risk of downtime," says Gary MacFadden, the founder of Parity Research.
Some of the most significant hard costs over the lifetime of the system include the following:
- Procurement costs
- Hardware and software maintenance costs including extended warranty costs
- Management staff costs during the life of the system
- Monitoring system costs
- Power, cooling and backup power supply costs
- Data center space costs
- Backup infrastructure costs
- Disaster recovery costs (including disaster recovery service subscriptions or maintaining compatible storage at secondary sites)
- Storage area networking costs
- Security costs
- End of life costs including the cost of migrating data to a new storage resource
In addition to the risk of unscheduled downtime, mentioned earlier, other soft costs that should be taken in to account include the following:
- Cost of waiting for storage to be provisioned
- Unused storage capacity cost
- Cost of storing duplicated data
- Scheduled outage costs
- Risk of lost data
- Poor storage performance costs
High Performance TCO
This last cost—the cost of poor storage performance—can be particularly significant in large enterprises with customer-facing applications, says MacFadden. "Some applications are all about serving customers, and if the screen doesn't refresh in 100ms the company could lose business worth $1m a day," he says.
Solid state drives (SSDs) offer far higher performance than HDDs, but they are also more expensive per gigabyte than HDDs. So if you are purchasing a storage array solely to provide extra storage capacity, the TCO of an SSD based array is bound to be higher if you assume that all the additional costs are about the same. (In actuality, you could argue that SSDs are more reliable and consume less energy, but let's ignore that for the moment.)
But if high performance storage is the main objective, then the economics, and the relative TCO, change dramatically. "If you are after high performance for a SQL app or an Exchange server, then you can use hard disk drives, but you need to buy many times more HDDs than SSDs to beef up the IOPS," says MacFadden. That's because enterprise grade SSDs may offer 4000 IOPS, while even the highest performance 15k rpm HDDs only offer around 400 IOPS. To get equivalent performance from an HDD-based array requires short stroking—a practice which only utilizes a small proportion of the capacity of a large number of drives.