Originally posted by: Kimmel
FlashSystem 900 is the core of our All-Flash Array series for distributed. It is the heart of the FlashSystem V9000, based on IBM Spectrum Virtualize code, and of the FlashSystem A9000, which uses the IBM Spectrum Accelerate/XIV code base to enable advanced functions, copy services and deduplication.
The new FlashSystem 900 AE3 model, which has been announced in October, is now generally available, building a new generation of our FlashSystem family. FlashSystem is an ultra-dense, ultra low-latency system, which now comes with three times the storage density compared to the previous generation in only 2U. This has been achieved with a new version of flash cells – IBM-enhanced 3D TLC (Triple-Level Cell) NAND flash technology.
As before, each FlashSystem 900 can host up to 12 MicroLatency flash modules, but these have much bigger capacities now, of 3.6 TB, 8.5 TB or 18 TB each. And, they have hardware-accelerated inline compression on top. FlashSystem compresses at all times without taking any performance away. Response times for small blocks are in the range of ~100 µs, with compression always on. While usable capacity per each 2U device can be up to 180 TB, compression savings enable even larger effective TB values. The IBM Comprestimator tool can help calculating FlashSystem compression savings. With IOps throughputs up to around 1 Million (4K), sequential throughput values of up to 10 GB/s (64K Reads) and the response times in the ~100 µs range, the new FlashSystem 900 AE3 has performance values comparable to its predecessor: but now with compression fully enabled throughout.
The SKLM external key manager for encryption is now also supported, with AES-256 encryption on the MicroLatency modules. The new FlashSystem 900 code also introduces a new user interface ("UI") with a similar look and feel like other recent IBM storage solutions, like the SVC V8.1 code base, for a consistent management experience across all platforms.
Connectivity is via Fibre Channel, or QDR InfiniBand.