Deploying solid-state drives (SSDs) in the data center is a great way to boost application responsiveness and remove I/O bottlenecks in the storage layer. But today’s data centers face a mixed bag when it comes to storage workloads, and the answer isn’t always as simple as “add more speed.”
There are different kinds of speed to consider, namely sequential read/write and random read/write. And what’s more, data centers also need the endurance to handle tens of thousands (potentially millions and billions) of I/O transactions daily. So when you’re looking for the right storage device for your data center, the best-fit solution really depends on the type of workload.
To help us sort through the options, let’s examine a few specific use cases in data centers today: retail databases, online analytical processing (OLAP), virtualized servers, and machine-generated data.
• Online retail sales databases help businesses deliver a seamless experience for customers while driving profitable e-commerce, tracking inventory, and forecasting sales trends. More and more consumers are making their purchases online, which ultimately drive demand for greater storage performance with heterogeneous read/write workloads.
• And then there’s big data analytics, which has been making waves in tech blogs for the past few years. Here’s a simplified description for the unacquainted: big data is the process of deciphering massive volumes of data in search of insights and previously unconsidered correlations that could lead to new efficiencies, process improvements, and even more competitive business models. Online analytical processing (OLAP) supports big data by enabling IT managers and data experts to answer multidimensional queries quickly and efficiently. Naturally, high volumes of data mean heavy workloads that put a lot of strain on storage.
• Let’s not forget cloud computing and the innovation that made it possible, virtualization. Virtualized servers run a single hypervisor that controls multiple simulated server environments, commonly referred to as virtual machines (VMs). IT managers can provision multiple VMs per physical server, making it possible to support multiple business units with varied needs. The challenge lies in staying nimble – the more VMs there are, the harder it is for storage to keep up.
• Finally, let’s consider machine-generated data. This is data that originates from any source without human intervention. It could derive from sensors in an electrical grid, manufacturing equipment on the factory floor, or log data from web servers and financial transactions.