I’m at Pure Accelerate 2019 trying to collect my thoughts and report on what I saw in the last couple of days. Some of the announcements were particularly exciting, others less so maybe, but I think that Pure Storage demonstrated, once again, to have a solid long-term strategy based on the idea of an “All-Flash Datacenter.” My only concern is that, in the short term, they could struggle a bit… but that is part of the game, isn’t it?
Last week I was wondering what to expect from Pure Accelerate this year, and my list included QLC flash support, more cloud, Intel Optane, and some info about what they were going to do about CompuVerde (a company they recently acquired). I have to say that they fulfilled all of my expectations and probably more:
- QLC: A new Flash Array // C has been launched. Based on QLC flash memory, with higher latency and capacity compared to other models. Up to 1.4PB raw and with the same functionalities of the rest of the FA family. This opens up new opportunities for Pure when capacity is more important than absolute performance. On paper, I didn’t get the technical deep dive that I was expecting, the general idea is that Pure worked to change how data is written in the backend (DirectFlash modules), limiting and optimizing write operations. The product is positioned for read-intensive workloads and relatively inactive data. FA//C can reach more than 5PB capacity after compression and deduplication, but my concern is that secondary data is often compressed and customers could take advantage of the capacity for IO intensive workloads like big data analytics, putting the array under considerable stress (hence, the modules could wear off quicker than projected). I’m pretty sure that this array will quickly become a best seller for Pure. I’m not as sure though about how it will really perform in real-world scenarios, but this is something we will discover soon.
- Intel Optane: A new DirectMemorymodule based on Intel Optane is now available. Contrary to the DirectFlash module, this one is used as a read-cache. Pure promises an overall speed bump on existing arrays that is pretty consistent, depending on the type of workloads of course. The good news is that this module will be available for everybody, both new and existing arrays, and can be installed in the same slots normally used for DrirectFlash modules. The only tradeoff is that you will sacrifice some capacity because you have to use one or more of these slots for these modules. But you can’t have everything, can you? And I love it when new features and options are retroactive.
- Cloud: Cloud Block Store is finally GA. I had the chance to see how it actually works and I can say that Pure has found a good compromise to build something that can give performance and availability in the cloud for its customers. It is still version 1.0 and there is room for improvement, but it is a good start. I think that in this area Pure is still behind the competition though and hope they will be able to fill the gap with additional products and functionalities soon.
- File and object (Compuverde): Nothing really announced here, but it is interesting to note that Pure is working to integrate Compueverde into or onto the FlashArray product family. We could expect something for 2020 with FA series that will be able to support files and object storage alongside block protocols.
- Pure-as-a-Service: Pure is working to improve its already good evergreen subscription program. Now every product from Pure can be purchased in this way. Not that this is something unique, every vendor is working on similar programs, but for its customers, this is a great opportunity to move more CAPEX to OPEX.
Pure has made other announcements, including enhancements to AIRI (their AI platform), VMware Analytics, and FlashBlade among others and you can find more about them on their website. I also suggest you check out Tech Field Day videos recorded here at Pure accelerate for technical details on most of the product announcements at Pure Accelerate.
Takeaways
The All-flash datacenter is not here yet (will it ever be?), but it is undeniable that Pure is executing on this long term goal with a solid strategy, balanced products, and services to support it.
Some gaps in the product line need to be filled somehow and soon, but it is clear that the company knows and is working on it. It is just a matter of time, but I’m sure they will be able to deliver what is necessary to enable their customers to embrace their hybrid cloud journey with the right tools.
It is unbelievable that this company is only 10 years old and it has been able to grow from being one of the many startups with an all-flash array to one of the primary vendors in data storage. Impressive. That said, I’d like to see more Data than Storage from Pure in the future. Data mobility and management are becoming fundamental parts of any conversation with enterprises and they have to come up with something credible on this soon.