Scott Buecker
Storage Engineer
Two weeks ago, a customer shared with me that his team experienced an encryption-related outage on their Tier 1 storage array. How could this happen? All we’re doing here is Data at Rest Encryption, right? And since when do we have outages on our Storage arrays? I thought that was over since most vendors figured out how to do non-disruptive code loads 15 years ago. Spinning disks are gone and Flash drive failures seem to be few and far between. What would you think was the most-likely culprit?
I expected an external key manager or power event that required manual recovery-key entry. He explained they had a requirement for host-based encryption. When enabled, the CPU and memory resources on their storage array went to 100% and performance cratered for all attached hosts. I won’t name names, but the customer was running a T1 Flash system that does all data reduction in software. It was obvious that the encryption change caused the problem, but they couldn’t disable data reduction at the volume level and could not afford to disable data reduction on the entire pool due to lack of real capacity. The only option was to roll back, start the RCA process and hang your head in defeat.
Most vendors in this space depend greatly on data reduction to be cost competitive. They also perform 100% of that data reduction in the storage controller software. If you’re not IBM, you’re basically a software company. Commodity TLC flash with a colorful logo and huge marketing budget… all you need is software to complete the solution.
What sets IBM Flash Storage Apart
IBM has a heritage of huge investments in R&D and engineering efforts to set them apart from other vendors. IBM has been evolving the FlashCore Module (FCM) since they acquired Texas Memory Systems in 2012. FCM utilizes the NVMe protocol, a PCIe Gen3/ Gen4 U.2 interface, and high-speed NAND memory to provide high throughput. The latest FCM being the v3 module which is available in the new IBM FlashSystem 9500 and 7300 arrays.
To understand why IBM Flash Storage is impressive, let’s do quick NAND flash refresher. Initially flash was delivered in the form of Single-Level Cell (SLC) and Multi-Level Cell (MLC) drives. SLC can store one bit per cell, and MLC can store two bits per cell. Most vendors today have landed on Triple-Level Cell (TLC), yes that’s three bits per cell, because going beyond that required too much housekeeping. As we progress from SLC to MLC to TLC, the capacity/density increases which has been critical to drive down the cost.
So why not just keep going, right? Let’s roll out Quad-Level Cell (QLC) so we can store four bits per cell and increase the capacity even further. The problem is that QLC introduced many issues related to managing hotspots, write amplification, garbage collection, health checks and error correction. Basically, we’ve hit the operational limit of Flash consumption with TLC. That’s still the case for vendors using commodity TLC NAND flash and managing this “housekeeping” in Storage Controller software.
Now consider the hardware IBM has added to each FlashCore v3 module; each FCM combines NAND flash, DRAM and MRAM for caching, and an FPGA compute element. With this horsepower IBM has managed to deliver a QLC-based flash-drive that is as reliable as TLC and often more performant. But this is just the beginning. The IBM FlashCore Module started with compression efficiencies and have continued to build intelligence into their Flash media at the hardware level.
Avoiding that Encryption Related Outage
Remember Company X who flipped the switch on Host-based encryption and took down all their Tier 1 applications? IBM has strategically positioned themselves, once again, with superior engineering to address these challenges quickly. And actually, this encryption scenario would likely not even impact the FS9500/FS7300 because the data reduction offload to these FCM monsters. And the excitement builds because of what’s coming next… Storage Class Memory, Managing Immutable Snapshots for Cyber Resiliency, and ultra-dense and cost-effective requirements to compete in this market.
Flash System Bundles
In response to our customers’ interest in sample IBM FlashSystem configurations and pricing, we recently created this Enterprise, Entry level IBM FlashSystem bundle sheet. As every environment is different, let us know if you’d like to see a different configuration.
Summary
Let’s be honest, IBM doesn’t have the “flashy” marketing with bright colors and celebrity spokesmen. They won’t try to sell you a Flash array that will last a decade, or strategically lock you into their storage. IBM just has the best Flash technology delivering enterprise-class, cost-efficient, high-performance storage solutions for today’s modern workloads, and I would encourage all storage systems consumers to challenge their current vendor to convince you otherwise.
More Information
Mainline is a Platinum level IBM Business Partner – the highest partnership level in the IBM PartnerWorld program. We have heavily invested in the technical skills and certifications necessary to provide the top level of services and solutions to our customers for all IBM hardware and software products, including maintenance and software support services. For further assistance with IBM Storage solutions contact your Account Executive directly or contact us here.
You may be interested in
BLOG: IBM FlashSystem FS7300 and FS9500 Cyber Resilience for Hybrid Clouds