The cost of storage is not just the price of purchasing physical hardware

Most storage managers agree that storage costs are not just the price of buying physical hardware. IT managers often add maintenance costs and the cost of running and managing storage for administrators. In fact, there are more components to the true cost of storage, and determining the true cost of storage systems requires determining the cost of x capacity to provide a level of Y-level performance to an application.

First let's look at the cost of providing capacity, for example, if you have an application that requires a net usable capacity of terabytes, how much storage do you actually need to buy to ensure this terabytes of storage? Obviously, you can't just buy 10 1 TB hard drives and then install them on the server and announce that the task is complete.

First, you can't run 100% of your capacity without impacting performance. We don't even see the capacity to run 30%, but let's say you're very efficient and storage utilization reaches 50%, which means that the terabytes of storage needs actually require terabytes of storage.

For reasonable assumptions, this application will require some form of data protection, RAID 5 or RAID 6. For RAID 5, we will need at least 16 hard drives, not 15, and the problem with RAID 5 is that the rebuild time is too long and you may be at risk of all data loss when rebuilding. So now you typically choose RAID 6,raid 6 with dual-disk parity technology, two disks for data protection, which means you'll need to buy 17 hard drives.

Next, you may need to save a copy of your data backup in case of a catastrophic raid failure or an application failure, and you need to revert to a previous copy of the data. For security purposes, you should keep a copy of your data backup at both local and remote sites.

Suppose at the remote site you want to use the storage system's replication capabilities. First of all, this means that you can't buy the cheapest storage system, at least a medium copy-capable system, which is an extra charge. This also means that the remote site's storage system needs to have functionality similar to the local storage system, and now you need to buy at least 34 drives, and two of them should be at least mid-range storage systems.

For local backups, you can select a slightly cheaper array or copy to disk. If you buy a cheap system instead of using the same array as the main application data, you can use a second array to prevent array failures caused by software or hardware. You can also select 2 TB drives to reduce overall costs.

In the second system you may still want to run RAID 6 because the 2 TB hard drive is rebuilt longer and the storage system is not strong, requiring a third system and 9 additional hard drives. Finally, you will need to install the software to make a copy, if your application has this feature, you can use the built-in features, otherwise you'll need to purchase the software.

To sum up, the most basic terabytes of storage requirements, with the most basic hardware components, require three systems and 43 hard drives on an application. We haven't even discussed whether these 1 TB hard drives can provide the performance you need. You may need more smaller, faster drives or solid-state drives to meet application performance requirements.

Finally, we need to consider some software features that affect capacity consumption, such as snapshots, automatic compact configuration, and duplicate data deletion. Over time, we will also need to cover performance and efficiency software and solutions to calculate the real cost of storage.

(Responsible editor: The good of the Legacy)