With the maturity of flash technology and huge real-time data demand, the high-level all-flash array has been popularized among enterprises and even SMB (Small and Medium Business). With the upgrade "cost", SMB is opting for low cost entry-level all-flash array. Therefore, more and more storage array vendors have launched the entry-level all-flash array, which strategically aims to develop different functions, so as to make the products rich and differential.
With so many choices, how to choose the solution that meets the needs of the current business?
Here is the 5P strategy you can't ignore :
Performance: IT application services are booming. It needs to solve the storage requirements of database and media streams, so all-flash array should have the ability to provide high IOPs and throughput with low latency.
Protocol: It is difficult to guarantee that enterprises are equipped with FC (Fiber Channel) and iSCSI (Internet Small Computer System Interface) network environment at the same time. The ability to support both protocols has become an important condition, which can reduce the cost of enterprise infrastructure construction and reduce external expenditure.
Persistence: As a data solution that consumes a large number of data immediately, uninterrupted flow and avoiding losing data are significant. Even a brief stop in working flow will cause a huge loss, so it's necessary to consider the high availability architecture design in the solution.
Price: The solution is not just the all-flash array but also SSD (Solid State Drive) that equip with. It should take consider the whole cost to purchase all-flash array for maximum benefit.
Position: Entry-level all-flash array is suitable to meet the high-performance needs of SMB, including, but not limited to, multiple databases, customized environments, and media-based service scenarios.
Performance vs. Software Features
Performance is undoubtedly the most important evaluation standard for an all-flash array. In the case of IOPS (Input/Output Operations Per Second), it can meet the high-performance requirement of the IT industry, but due to the high storage space price per unit, in addition to the performance, some IT personnel expect to have the data deduplication and compression high-end features. But if IT personnel add the features to perform tasks at the entry-level all-flash array, it will drop the performance hard.
Why does the performance decrease when high-end functions are added?
According to the test data, performing data deduplication and compression consume massive processor resource requirements. Enabling compression decreases performance by more than 50% and results in higher extra cost, so it is necessary to discard the feature in the case of a limited budget.
We can use the following data to understand that after the implementation of data deduplication/compression.
All-Flash Array's data under different conditions :
Random Read 4K :
After Enabling compression/data deduplication, IOPS dropped 9.5%
Random Write 4K：
After Enabling compression/data deduplication, IOPS dropped 65.5%。
Sequential Read 64K：
After Enabling compression/data deduplication, throughput dropped 68.5%。
Sequential Write 64K：
After Enabling compression/data deduplication, throughput dropped 69.5%。
We can see the impact after enabling the compression/data deduplication feature. Random write 4K and throughput drop to 1/3 of the original performance.
Considering if we can complete the thirty tasks in a day, 10 tasks can be completed after the features enabled. The time consumption increased up to three times, it will have a high cost of time and labor, and may even lose important business opportunities.
Looking back at the 5P strategy in the first paragraph.
It's important that the all-flash array supports iSCSI and fiber channel protocols. With protocol flexibility, SMB can choose the protocol according to its environment.
All-flash array is becoming a key role in the IT environment. The persistence is important to keep the all-flash array stable. Considering if unexpected errors occur during operation, which results in system shutdown or even data loss, it could be an unaffordable loss. With the high availability architecture design, the all-flash array can provide continuous services in the event of a single controller failure, and doesn't affect businesses and services due to failure. The high availability architecture design is also convenient in maintenance, IT personnel only need to plug out the fault controller and plug in the new one in when the controller failure occurs.
Applying the all-flash array at an appropriate price is important. You should aware that the price includes both the all-flash array and the attached SSDs, you need to be careful to set up a high-performance all-flash array with an affordable budget.
The entry-level all-flash array is positioned to meet the high computing requirements for SMB. The suitable application scenario includes the database, virtualization environment, media entertainment, e-commerce, etc....
Let's return to the title "Choosing the right entry-level all-flash array". The all-flash array must be selected for performance reasons, but IT personnel often want to have data deduplication/compression high-end features at the same time.
Through the chart display, it's clear that the high-end features of the entry-level all-flash array cause a sharp decline in performance.
Therefore, the entry-level all-flash array is not suitable for high-end features. The 5P strategy can make your purchase easier for the highest performance at a limited cost.
If you want to have an all-flash array with the high-end feature, the medium/high-level all-flash array that has higher processor performance will be a better choice. But be careful that it may exceed a lot of budgets.