Three new IBM storage systems based on IBM FlashSystem

Colleagues, our company presents three new data storage systems based on IBM FlashSystem technology. First of all, these systems are designed for business - medium and large, as well as cloud providers. New systems can significantly improve performance and reduce the load on virtual machines, which in many cases is very useful.

Three new models of solid-state arrays are IBM FlashSystemA9000, IBM FlashSystemA9000R, and DS8888. The first two models, IBM FlashSystemA9000 and IBM FlashSystemA9000R, are designed for cloud providers. We recommend using them to quickly and efficiently deploy hyper-scalable private clouds. As for the third model, DS8888, this is a corporate system for storing z-Systems mainframe data (more about them here ) and IBM PowerSystems based on Unix. An updated line of storage systems was prepared with Sandisk.

With high performance and reliability, these solid-state arrays will help ensure the smooth operation of critical business applications, as well as cloud and cloud applications. Flash technology, on the basis of which all this is created, reduces the time to work with BigData.
')

IBM FlashSystem A9000 and A9000R

Let's start with the IBM FlashSystemA9000 model. This out-of-the-box solution is fully configured. The system works in a variety of cloud environments, with the inclusion of the compression function in real-time mode due to the removal of intermediate stages such as templates and deduplication. The model configuration includes from 60 to 300 TB of memory, but, if there is a desire, the volume can be increased to 1200 TB.



The architecture allows you to scale storage up to 144 nodes with a base in the form of two 8-core IntelXeonE5 v3 processors with a frequency of 2.4 GHz for each. Three controllers are installed here, as well as a FlashEnclosure unit. As for this block, it consists immediately of 12 individual MicroLatency modules. The capacity of each can be 1.2, 2.9 and 5.7 TB. IBMFlashCore provides low latency with a rate not higher than 250 ms.

The model should be used by a medium or large business, where there is a need for storing and processing large amounts of data, as well as the smooth operation of cloud applications and the clouds themselves.

Model A9000R is the older relative of the A9000. The principle of operation is the same, but the performance is even higher. So, the architecture here is four-controller + 2 FlashEnclosure units. Scaling to a maximum of 12 controllers and 6 blocks is also possible. The developers have foreseen the presence of InfiniBand switches. Storage capacity - from 300 to 1800 TB with the highest possible rate of 3000 TB. Note that this system requires a whole rack.



Both models work with data deduplication, which allows you to compress data five times. According to our calculations, the cost of storing information is $ 1.5 per GB. The memory type is IBMMLC (Multi-LevelCell). The system performance is about 500 thousand IOPS with a minimum delay of 250 µs.

And now - IBMDS8888

These are the so-called all-flash-systems designed for enterprise-class servers. The owner of such a unit can improve the performance of software solutions with intensive data and application processing. Accordingly, the productivity and speed of work of most business processes, which depend on the speed of information processing, are increased.

For user convenience, the system is equipped with a Hyper-ScaleManager interface, which greatly simplifies the task of moving data and managing hundreds of nodes with just one tool. Experts estimate that Hyper-ScaleManager roughly halves the number of mouse clicks required for an operator to work.

At the moment, the fastest and most efficient storage systems are those systems that use flash memory, which allows to consume less energy. Such equipment is also more productive, and takes up less space. The use of such technology speeds up the processing of large data arrays, reduces the cost of storing information per 1 GB and reduces the time it takes to work with company data.