Solid state storage not only changes the storage industry, but also affects the entire computer field. We have seen that flash memory has completely changed the consumer electronics field, replacing rotating disk drives in almost every category of consumer electronics.
This enthusiasm is also extending to the data center. Database Administrators, System administrators, and application system owners have begun to realize the benefits of solid state storage. They see the advantages of solid state storage in terms of performance and energy consumption, but still have doubts about the durability of enterprise-level applications. However, the storage industry is proposing a new flash controller that can improve the service life and performance of low-cost Flash Media to replace expensive enterprise-level Flash Media in enterprise-level application environments.
Best SSD Application Method
SSD) provides a feasible and faster solution to replace the traditional HDD. The first step in selecting the appropriate storage is to clarify the storage performance requirements of applications. This demand will determine whether it is solid state storage or traditional hard disk storage is a more suitable and cost-effective solution.
We start with a relatively simple example and gradually extend it to complex situations. SSD is already involved in a data center laptop. The SSD of the laptop provides Fast startup and comprehensive performance, and significantly increases the battery standby time because the built-in SSD consumes very little power. Some applications, such as word processing, large image files, macro tables, databases, and videos, can respond very quickly when SSD is used. The speed for copying files between laptops is also very fast. Therefore, the effect of SSD is not only reflected in the performance improvement of applications.
The same benefits can be obtained for desktop PCs. For example, using solid state storage to start a computer is an easy and inexpensive way to improve performance. Using SSD as the boot disk on a desktop computer, I/O acceleration can extend the service life of an old machine. This is also applicable to the old laptop, as long as you have the appropriate SSD interface.
Accelerate database operations
Any application that needs to improve performance or reduce storage access latency is a good use of solid state storage technology. For example, many database operations are actually a combination of many small requests, such as table scan and query operations, which are executed in order, the output result of one request is the input of another request. The final result is returned to the application after all the requests are completed. In this case, the use of solid state storage significantly reduces latency and faster turnaround), bringing a huge difference to the overall performance and user experience of applications.
The best enterprise hard drive has an average tracing time of about 2 ms for each request, and not every storage system uses high-speed cache due to data protection. Therefore, even if you can obtain the same performance as the hard drive, using SSD will provide better overall latency because they have no track time at all. Imagine that when a large number of database transactions are running, every I/O encounters a tracing delay for the enterprise hard disk. Imagine that there is no tracing delay for a large number of complex database operations, faster storage devices, you will find that this is why SSD is so good for database applications.
SSD layering
The performance obtained by Solid storage technology is not limited to database applications. This is why we have seen the increase in cache and layered solutions. Most servers, whether independent or cluster, are under great pressure due to a large number of application jobs. Each server has different busy and idle time. If you store all your data in solid state storage, you don't need to consider caching and layering. But if your data center is like most data centers, this technology should be considered if most of the data of the current application is stored on the traditional rotating hard drive.
With SSD Tiering technology, users can decide when data is stored on Solid State disks. Layers can be executed manually, or automatically by using layer software on the host or storage controller. The layering technology will migrate specific hotspot data to the SSD layer at the right time, and migrate them back to the low-speed Disk layer at the right time. If the hierarchy is manually executed, the Administrator must constantly observe the I/O activities to determine when to migrate the files or data. You have to manually track the Access frequency of each file in your system to determine when the data will be migrated out of the SSD Layer Based on the access. For systems of different sizes, manual execution is almost impossible, which requires automatic layering of software. With Automatic hierarchical software, access to files and data is automatically tracked, and data migration is executed at a predetermined time according to user-defined policies. The adoption of layered technology has benefited applications that have migrated data to a fast level, and the performance improvement is immediate and significant. If you have multiple applications that need to improve performance, but you cannot choose or have no time to test which applications need to improve performance), the automatic layering solution is a good choice. If you only have one application to benefit from the layering technology, you do not need the automatic layering software. Most data centers with dozens, hundreds, or even thousands of application systems can use this technology to achieve higher performance.
Solid state cache technology
Another technology is SSD cache. Data cache to SSD is determined by the host software or storage controller. This technology copies the data to the SSD cache, you do not need to migrate data from the original location determined by the user and application software. Cache Technology is very easy to manage, because all decisions are made by the cache software or controller. The cache technology calls hotspot data into the cache for access to accelerate application access, and the performance is gradually improved as more data enters the cache. We call this step-by-step performance improvement "warm-up" or "Climb", which takes several minutes or hours, depending on the number of devices and I/O operations. Depending on the device, the cached data can be read-only or read/write. SSD Cache Technology follows many cache algorithms that are the same as memory cache technology, or even the same as the internal Cache Technology of the processor. Some SSD cache solutions not only cache the obvious hotspot data, but also obtain the adjacent data in advance. The cache Software determines that this part of data will become a hotspot according to its I/O mode. Most cache solutions allow administrators to decide which files or volumes can enter the cache to improve performance, so you can exclude specific data to prevent it from blocking the cache. If you think that most or all applications can improve performance, you should consider using SSD cache technology.
With the layered and cache technology, SSD can improve the performance of more capacity. For these two solutions, you must calculate the required SSD capacity. In most cases, it is considered that only 3% of the total disk storage space or up to 10% of the SSD capacity can be significantly improved.
SSD data compression
Compression technology is another new topic because of solid state storage technology. Because SSD technology is generally more expensive than traditional hard drive technology, while only paying attention to the price per GB, one way to improve the benefits of SSD technology is to compress data before it is stored in SSD, thus consuming less valuable SSD space resources. At present, with the performance improvement of the processor, some additional CPU processing capabilities are used to compress data and then store the data on a fast storage device, thus improving the overall performance. This requires consideration of whether compression occurs on the host or on the storage system. Some storage can be compressed, and the data is immediately compressed when it enters the storage system. All cached, SSD, or disk devices can only see the compressed data, in this way, you can use less capacity.
Another solution, as mentioned above, is the all-flash array. This will become more common, with large storage vendors starting to follow up on the trend guided by some startups last year. We can expect to see the release of its flash array from a large storage vendor this year. It is likely that in the near future, the all-flash array will have the same advanced features as the current hard drive system, including features such as automatic streamlining, deduplication, and more.