OK, you got me. If you are buying Oracle RAC licenses, it is already not cheap, but what if you need RAC’s high availability and scalability? Is there a way to minimize the hardware costs, yet maintain the extremely high performance of a full- or multi-rack based server along with the complimentary SAN switches and SAN storage arrays? There is a way.
Most Oracle DBAs are familiar with the parameter cache hit ratio, a metric that monitors the rate at which Oracle finds the data blocks it needs in memory over the lifetime of an instance. Typically a well-tuned OLTP would have a range of 90 to 98 percent cache hit ratio and a well-tuned Oracle warehouse database would have a 60 percent cache hit ratio. The lower cache hit ratio for warehouse databases is indicative of the fact that the amount of data coming into the database buffer cache could very well overwhelm the cache resulting in more data being accessed directly from the storage device (typically disk).
Whether you are a social network or a web-based retailer, the responsiveness of your website is at the core of the business. If a customer is unable to get to the information they are seeking, or purchase the item they desire quickly, they will find another alterative that is less frustrating. To address this need for rapid response time, web scale organizations are frequently turning to flash storage as the most effective solution.
Even with the recent drop in prices narrowing the gap between flash storage and hard disk drives (HDDs), flash takes a lot of heat for still being too expensive. But if flash is measured on its ability to deliver high performance resulting in improved response times and lower latencies, is it fair to base price comparison solely on capacity? The short answer is no. Let’s look at some alternative ways to more accurately analyze the cost of flash.
While it’s widely accepted that solid-state drives (SSDs) provide significant performance advantages over hard disk drives (HDDs), there are three main related factors to consider when deploying SSDs: form factor, performance, and location in the storage tier or in the server. To help customers understand the trade-offs between the form factor, performance, and location of various types of SSDs, Virident just published a white paper, “The Advantages of Server-Side Flash Storage.”
In my last post I wrote that while flash was introduced at the right time to help manage multi-core CPUs, big data, and server sprawl, there are various grades of flash storage that one must know about when considering its use. In this second post I’ll address the realities of various categories of flash storage, their performance variability, and the real-world deployment motivations that separate consumer-grade from enterprise-grade flash solutions.