FlashArray: //C All-QLC-NVMe-Array with even more flexibility ... also with a little brother.
[ NOTE: machine translation with the help of DeepL translator without additional proofreading and spell checking ]
FlashArray //C has already been an integral part of the FlashArray product family for about one and a half years and is enjoying increasing customer demand. Especially customers who prefer homogeneous infrastructures for different (compared performance- or capacity-oriented) workloads show special interest in FlashArray //C. The many advantages of NVMe have already become widely known and many pain points in the use of obsolete storage technologies have become obsolete as a result. I personally see more and more companies underestimating their growth and looking for short-term alternatives to accommodate their exploding data in a flexible and cost-efficient way. Especially in virtual infrastructures, on-board features make it quite easy to automate storage relocation of individual applications based on policies. Applications with large capacities but fewer performance requirements can thus be placed on less expensive NVMe QLC. The virtualization level of companies is quite high, so I am looking for an application example here accordingly, but the above requirements can also be found at any time with physical servers (especially e.g. file servers7 volumes).
A storage system optimized to accommodate large capacities and with comparable performance of a FlashArray //X based on new technologies is required here today. Hybrid storage systems with a mixture of different SAS/SATA technologies are out nowadays!
FlashArray //C was built exactly for these requirements and has been continuously developed since its market release. I already reported in detail on FlashArray //C in September 2019 in the article "FlashArray: this is the brand new FlashArray family member - FlashArray //C - a deep dive".
Categorization by application of FlashArray systems is individual and varies according to customer requirements. You can outsource Tier 2 systems, test/dev systems e.g. snapshots to FlashArray //C. A lot of different scenarios are possible.
Now there are more innovations for FlashArray //C! ...
The innovations
FlashArray //C60 gets a refresh of the system components and will be available as //C60 R3 in the future. In addition, an entry-level //C40 R3 system has been developed specifically for mid-market, ROBO sites or even DR sites with lower capacity requirements.
FlashArray //C40 scales from 247TB RAW capacity up to 494TB and effectively provides 1.8PB of storage on 3 height units at a DR of 5:1. As usual, a controller upgrade can be performed from C40 to C60, after which the maximum allowable RAW capacity would be increased to 1877TB RAW by connecting additional DirectFlashShelves. This would mean 7.2PB effective usable capacity (at 5:1) on 9 height units.
It is not possible to run R3 shelves on a //C40 as of today. An inter-chassis controller swap from //C to //X or vice versa is not possible.
Module sizes of 24.7TB or 49.2TB QLC NAND so-called DFMc's are available - mixed operation with the module sizes is not possible. FlashArray //C R3 requires Purity 6.1.
Since FlashArray //C R3 it is now also possible to scale in steps of two DirectFlash modules (c). The storage principles: min. 10x modules in the chassis and 14x modules in the shelf still apply. A small difference between //C (QLC) and //X (TLC): a different RAID scheme is used for QLC modules. With QLC the RAID overhead is mapped with N+2+1 instead of N+2+2 for write groups. This makes a QLC configuration a less effective one and allows them multiple configuration options.
Conceivable extensions can be represented as follows (confirmed):
Both //C40 and //C60 support unified operation with file and block.
This is not true!
Rumors are spread that QLC modules are unreliable ...
Pure Storage DFMs and the associated firmware (DirectFlashSoftware) are a complete in-house development. The granular control of the QLC-based DFMs is done through the DirectFlashSoftware, this means that data management (writes and reads), data placement and background task scheduling is specific to the QLC NAND geometry. This gives the storage (//C and //X) a significant improvement in read performance, eliminating the need for wear-leveling software to improve module lifetime and overprovisioning storage techniques to gain additional capacity. This gives Pure Storage's modules a significant advantage over the QLC drives associated with commercial quality.
More info - Links
All officially published setting options in the GUI but also CLI can be read via the "on-board" user guides of the Pure Storage systems.
Click on "Help" in the Purity main menu.
The User Guide is structured like the main menu and can be opened downwards. A search function is also integrated - within here you can also search for keywords.
WEB: Pure Storage (Pure1) support portal - Ticket system and support *(requires registered FlashSystems)
PHONE: Pure Storage phone support: GER - (+49) (0)800 7239467; INTERNATIONAL - (+1) 650 7294088
WEB: Pure Storage OFFICIAL blog